Publications resulting from research conducted using Delta appear here. Check back to see how the list of exciting discoveries made using Delta grows.
If you have a publication that should be listed here and isn’t, please share your success with us!
5005740
XHSH9DGT
1
nature
50
default
1693
https://delta.ncsa.illinois.edu/wp-content/plugins/zotpress/
%7B%22status%22%3A%22success%22%2C%22updateneeded%22%3Afalse%2C%22instance%22%3Afalse%2C%22meta%22%3A%7B%22request_last%22%3A600%2C%22request_next%22%3A50%2C%22used_cache%22%3Atrue%7D%2C%22data%22%3A%5B%7B%22key%22%3A%223P7WPAF4%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Hsu%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHsu%2C%20H.-Y.%2C%20Cheng%2C%20T.%2C%20Wen%2C%20J.%2C%20Schwing%2C%20A.%20G.%20%26amp%3B%20Wang%2C%20S.%20Seeing%20Without%20Eyes%3A%204D%20Human-Scene%20Understanding%20from%20Wearable%20IMUs.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.21926%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.21926%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Seeing%20Without%20Eyes%3A%204D%20Human-Scene%20Understanding%20from%20Wearable%20IMUs%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hao-Yu%22%2C%22lastName%22%3A%22Hsu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tianhang%22%2C%22lastName%22%3A%22Cheng%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jing%22%2C%22lastName%22%3A%22Wen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alexander%20G.%22%2C%22lastName%22%3A%22Schwing%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shenlong%22%2C%22lastName%22%3A%22Wang%22%7D%5D%2C%22abstractNote%22%3A%22Understanding%20human%20activities%20and%20their%20surrounding%20environments%20typically%20relies%20on%20visual%20perception%2C%20yet%20cameras%20pose%20persistent%20challenges%20in%20privacy%2C%20safety%2C%20energy%20efficiency%2C%20and%20scalability.%20We%20explore%20an%20alternative%3A%204D%20perception%20without%20vision.%20Its%20goal%20is%20to%20reconstruct%20human%20motion%20and%203D%20scene%20layouts%20purely%20from%20everyday%20wearable%20sensors.%20For%20this%20we%20introduce%20IMU-to-4D%2C%20a%20framework%20that%20repurposes%20large%20language%20models%20for%20non-visual%20spatiotemporal%20understanding%20of%20human-scene%20dynamics.%20IMU-to-4D%20uses%20data%20from%20a%20few%20inertial%20sensors%20from%20earbuds%2C%20watches%2C%20or%20smartphones%20and%20predicts%20detailed%204D%20human%20motion%20together%20with%20coarse%20scene%20structure.%20Experiments%20across%20diverse%20human-scene%20datasets%20show%20that%20IMU-to-4D%20yields%20more%20coherent%20and%20temporally%20stable%20results%20than%20SoTA%20cascaded%20pipelines%2C%20suggesting%20wearable%20motion%20sensors%20alone%20can%20support%20rich%204D%20understanding.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.21926%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.21926%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-22T13%3A13%3A40Z%22%7D%7D%2C%7B%22key%22%3A%22ST2964CK%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wang%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWang%2C%20D.%2C%20Zhang%2C%20Y.%2C%20Qin%2C%20Y.%2C%20Shao%2C%20Q.%20%26amp%3B%20Xu%2C%20D.%20Efficient%20Adaptation%20of%20Structure%26%23x2010%3BAware%20Protein%20Language%20Models%20for%20Diverse%20Protein%20Applications.%20%26lt%3Bi%26gt%3BCurrent%20Protocols%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B6%26lt%3B%5C%2Fb%26gt%3B%2C%20e70377%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Efficient%20Adaptation%20of%20Structure%5Cu2010Aware%20Protein%20Language%20Models%20for%20Diverse%20Protein%20Applications%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Duolin%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yichuan%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yongfang%22%2C%22lastName%22%3A%22Qin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Qing%22%2C%22lastName%22%3A%22Shao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Dong%22%2C%22lastName%22%3A%22Xu%22%7D%5D%2C%22abstractNote%22%3A%22Abstract%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20Protein%20language%20models%20%28PLMs%29%20trained%20solely%20on%20sequence%20data%20have%20significantly%20advanced%20our%20understanding%20of%20protein%20biology%20and%20achieved%20remarkable%20performance%20in%20protein%20prediction%20tasks.%20However%2C%20their%20lack%20of%20three%5Cu2010dimensional%20%283D%29%20structural%20features%20limits%20their%20predictive%20power%20in%20applications%20that%20rely%20heavily%20on%203D%20conformation.%20To%20address%20this%20limitation%2C%20we%20developed%20two%20structure%5Cu2010aware%20PLMs%2C%20S%5Cu2010PLM1%20and%20S%5Cu2010PLM2%2C%20that%20employ%20multi%5Cu2010view%20contrastive%20learning%20to%20align%20protein%20sequences%20with%20their%203D%20structures%20in%20a%20unified%20latent%20space.%20S%5Cu2010PLM1%20represents%20structural%20information%20using%20contact%20maps%20encoded%20by%20a%20pretrained%20Swin%5Cu2010Transformer%2C%20while%20S%5Cu2010PLM2%20directly%20encodes%203D%20backbone%20coordinates%20through%20a%20Geometric%20Vector%20Perceptron%20%28GVP%29%5Cu2010based%20model.%20The%20paired%20sequence%5Cu2010structure%20data%20were%20obtained%20from%20AlphaFoldDB.%20For%20both%20models%2C%20we%20designed%20efficient%20tuning%20strategies%20that%20enable%20optimal%20performance%20with%20minimal%20computational%20cost.%20Here%2C%20we%20present%20detailed%20protocols%20for%20adapting%20S%5Cu2010PLM1%20and%20S%5Cu2010PLM2%20for%20diverse%20protein%20applications.%20The%20protocols%20provide%20step%5Cu2010by%5Cu2010step%20guidance%20on%20generating%20structure%5Cu2010aware%20representations%20from%20S%5Cu2010PLMs%2C%20fine%5Cu2010tuning%20them%20for%20various%20protein%20prediction%20tasks%2C%20and%20using%20S%5Cu2010PLM2%20to%20produce%20structure%20embeddings%20for%20structure%5Cu2010based%20downstream%20analyses.%20We%20also%20provide%20source%20code%20and%20Google%20Colab%20implementations%20for%20easy%20customization%20and%20deployment.%20%5Cu00a9%202026%20Wiley%20Periodicals%20LLC.%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20Basic%20Protocol%201%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%3A%20Generating%20structure%5Cu2010aware%20representations%20of%20protein%20sequences%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20Basic%20Protocol%202%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%3A%20Efficient%20tuning%20of%20structure%5Cu2010aware%20protein%20language%20models%20for%20diverse%20protein%20applications%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20Basic%20Protocol%203%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%3A%20Using%20S%5Cu2010PLM2%20to%20generate%20protein%20structure%20representations%20and%20conduct%20structure%5Cu2010based%20clustering%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20Support%20Protocol%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%3A%20Google%20Colab%20quick%20start%20notebooks%22%2C%22date%22%3A%2205%5C%2F2026%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1002%5C%2Fcpz1.70377%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fcurrentprotocols.onlinelibrary.wiley.com%5C%2Fdoi%5C%2F10.1002%5C%2Fcpz1.70377%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222691-1299%2C%202691-1299%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-13T18%3A30%3A28Z%22%7D%7D%2C%7B%22key%22%3A%22GQRIQ8HL%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lee%20et%20al.%22%2C%22parsedDate%22%3A%222026-05-06%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLee%2C%20W.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Core%5C%2FShell%20HgCdTe%5C%2FHgCdSe%20Quantum%20Dots%20for%20Wavefunction%20Engineering%20with%20Infrared%20Band%20Gaps.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.26434%5C%2Fchemrxiv.15002806%5C%2Fv1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.26434%5C%2Fchemrxiv.15002806%5C%2Fv1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Core%5C%2FShell%20HgCdTe%5C%2FHgCdSe%20Quantum%20Dots%20for%20Wavefunction%20Engineering%20with%20Infrared%20Band%20Gaps%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Wonseok%22%2C%22lastName%22%3A%22Lee%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Erick%20I.%20Hernandez%22%2C%22lastName%22%3A%22Alvarez%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Madeleine%20J.%22%2C%22lastName%22%3A%22Fort%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Janice%20E.%22%2C%22lastName%22%3A%22Boercker%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andr%5Cu00e9%22%2C%22lastName%22%3A%22Schleife%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andrew%20M.%22%2C%22lastName%22%3A%22Smith%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026-05-06%22%2C%22DOI%22%3A%2210.26434%5C%2Fchemrxiv.15002806%5C%2Fv1%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fchemrxiv.org%5C%2Fdoi%5C%2Ffull%5C%2F10.26434%5C%2Fchemrxiv.15002806%5C%2Fv1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T18%3A27%3A16Z%22%7D%7D%2C%7B%22key%22%3A%22JBGDZEXM%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Mehdi%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMehdi%2C%20S.%2C%20Cho%2C%20I.%20%26amp%3B%20Isayev%2C%20O.%20Knowing%20when%20to%20trust%20machine-learned%20interatomic%20potentials.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2605.00640%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2605.00640%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Knowing%20when%20to%20trust%20machine-learned%20interatomic%20potentials%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shams%22%2C%22lastName%22%3A%22Mehdi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ilkwon%22%2C%22lastName%22%3A%22Cho%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Olexandr%22%2C%22lastName%22%3A%22Isayev%22%7D%5D%2C%22abstractNote%22%3A%22Prevailing%20machine-learned%20interatomic%20potential%20%28MLIP%29%20uncertainty-quantification%20methods%20rely%20on%20ensembles%20of%20independently%20trained%20backbones.%20These%20methods%20scale%20unfavorably%20with%20foundation-scale%20MLIPs%2C%20and%20their%20member-disagreement%20signals%20correlate%20weakly%20with%20per-molecule%20prediction%20error.%20Here%20we%20probe%20the%20frozen%20per-atom%20representations%20of%20a%20pretrained%20MLIP%20with%20a%20compact%20discriminative%20classifier%2C%20recasting%20MLIP%20uncertainty%20quantification%20as%20selective%20classification%20rather%20than%20error%20regression.%20The%20resulting%20method%2C%20PROBE%20%28Post-hoc%20Reliability%20frOm%20Backbone%20Embeddings%29%2C%20produces%20a%20per-prediction%20reliability%20probability%20that%20monotonically%20tracks%20actual%20error%20without%20modification%20to%20the%20underlying%20model.%20Across%20large%20held-out%20evaluation%20sets%20and%20two%20structurally%20distinct%20MLIP%20architectures%2C%20PROBE%20outperforms%20ensemble%20disagreement%20as%20a%20binary%20reliability%20signal%2C%20which%20strengthens%20with%20the%20expressiveness%20of%20the%20backbone%20representation%2C%20implying%20a%20favorable%20scaling%20trajectory%20toward%20foundation-scale%20MLIPs.%20Multi-head%20self-attention%20additionally%20yields%20per-atom%20importance%20maps%2C%20providing%20chemically%20interpretable%20diagnostics%20at%20no%20additional%20computational%20cost.%20PROBE%20is%20post-hoc%20and%20architecture-agnostic%2C%20and%20is%20directly%20deployable%20on%20any%20MLIP%20that%20exposes%20per-atom%20representations.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2605.00640%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.00640%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T17%3A14%3A49Z%22%7D%7D%2C%7B%22key%22%3A%22IWD84CDN%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Chan%20et%20al.%22%2C%22parsedDate%22%3A%222026-05-01%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BChan%2C%20J.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20ReLay%3A%20Personalized%20LLM-Generated%20Plain-Language%20Summaries%20for%20Better%20Understanding%2C%20but%20at%20What%20Cost%3F%20%26lt%3Bi%26gt%3BarXiv.org%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.00468v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.00468v1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22ReLay%3A%20Personalized%20LLM-Generated%20Plain-Language%20Summaries%20for%20Better%20Understanding%2C%20but%20at%20What%20Cost%3F%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Joey%22%2C%22lastName%22%3A%22Chan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yikun%22%2C%22lastName%22%3A%22Han%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jingyuan%22%2C%22lastName%22%3A%22Chen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Samuel%22%2C%22lastName%22%3A%22Fang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lauren%20D.%22%2C%22lastName%22%3A%22Gryboski%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alexandra%22%2C%22lastName%22%3A%22Lee%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sheel%22%2C%22lastName%22%3A%22Tanna%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Qingqing%22%2C%22lastName%22%3A%22Zhu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhiyong%22%2C%22lastName%22%3A%22Lu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lucy%20Lu%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yue%22%2C%22lastName%22%3A%22Guo%22%7D%5D%2C%22abstractNote%22%3A%22Plain%20Language%20Summaries%20%28PLS%29%20aim%20to%20make%20research%20accessible%20to%20lay%20readers%2C%20but%20they%20are%20typically%20written%20in%20a%20one-size-fits-all%20style%20that%20ignores%20differences%20in%20readers%26%23039%3B%20information%20needs%20and%20comprehension.%20In%20health%20contexts%2C%20this%20limitation%20is%20particularly%20important%20because%20misunderstanding%20scientific%20information%20can%20affect%20real-world%20decisions.%20Large%20language%20models%20%28LLMs%29%20offer%20new%20opportunities%20for%20personalizing%20PLS%2C%20but%20it%20remains%20unclear%20whether%20personalization%20helps%2C%20which%20strategies%20are%20most%20effective%2C%20and%20how%20to%20balance%20personalization%20with%20safety.%20We%20introduce%20ReLay%2C%20a%20dataset%20of%20300%20participant--PLS%20pairs%20from%2050%20lay%20participants%20in%20both%20static%20%28expert-written%29%20and%20interactive%20%28LLM-personalized%29%20settings.%20ReLay%20includes%20user%20characteristics%2C%20health%20information%20needs%2C%20information-seeking%20behavior%2C%20comprehension%20outcomes%2C%20interaction%20logs%2C%20and%20quality%20ratings.%20We%20use%20ReLay%20to%20evaluate%20five%20LLMs%20across%20two%20personalization%20methods.%20Personalization%20improves%20comprehension%20and%20perceived%20quality%2C%20but%20it%20also%20raises%20the%20risk%20of%20reinforcing%20user%20biases%20and%20introducing%20hallucinations%2C%20revealing%20a%20trade-off%20between%20personalization%20and%20safety.%20These%20findings%20highlight%20the%20need%20for%20personalization%20methods%20that%20are%20both%20effective%20and%20trustworthy%20for%20diverse%20lay%20audiences.%22%2C%22date%22%3A%222026%5C%2F05%5C%2F01%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.00468v1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T17%3A07%3A57Z%22%7D%7D%2C%7B%22key%22%3A%22D5PMEPJZ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Yarici%20and%20AlRegib%22%2C%22parsedDate%22%3A%222026-05-03%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BYarici%2C%20Y.%20%26amp%3B%20AlRegib%2C%20G.%20MER-DG%3A%20Modality-Entropy%20Regularization%20for%20Multimodal%20Domain%20Generalization.%20%26lt%3Bi%26gt%3BarXiv.org%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.01967v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.01967v1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22MER-DG%3A%20Modality-Entropy%20Regularization%20for%20Multimodal%20Domain%20Generalization%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yavuz%22%2C%22lastName%22%3A%22Yarici%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ghassan%22%2C%22lastName%22%3A%22AlRegib%22%7D%5D%2C%22abstractNote%22%3A%22Deploying%20multimodal%20models%20in%20real-world%20scenarios%20requires%20generalization%20to%20new%20environments%20where%20recording%20conditions%20differ%20from%20training%2C%20a%20challenge%20known%20as%20multimodal%20domain%20generalization%20%28MMDG%29.%20Standard%20architectures%20employ%20separate%20encoders%20for%20each%20modality%20and%20a%20fusion%20module%2C%20training%20the%20system%20end-to-end%20by%20optimizing%20on%20the%20fused%20features.%20In%20this%20paper%2C%20we%20identify%20that%20such%20joint%20optimization%20causes%20encoders%20to%20exploit%20cross-modal%20co-occurrences%2C%20statistical%20relationships%20between%20modalities%20that%20arise%20from%20source-specific%20recording%20conditions%2C%20rather%20than%20learning%20domain-invariant%20features.%20We%20term%20this%20failure%20mode%20Fusion%20Overfitting.%20To%20address%20this%2C%20we%20propose%20Modality-Entropy%20Regularization%20for%20Domain%20Generalization%20%28MER-DG%29%2C%20which%20maximizes%20the%20entropy%20of%20each%20encoder%26%23039%3Bs%20feature%20distribution%20to%20preserve%20feature%20diversity.%20MER-DG%20is%20architecture-agnostic%20and%20integrates%20into%20existing%20multimodal%20frameworks%20as%20an%20additive%20loss%20term.%20Extensive%20experiments%20on%20EPIC-Kitchens%20and%20HAC%20benchmarks%20demonstrate%20average%20improvements%20of%20approximately%205%25%20over%20standard%20fusion%20and%20approximately%202%25%20over%20state-of-the-art%20methods.%22%2C%22date%22%3A%222026%5C%2F05%5C%2F03%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.01967v1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T17%3A01%3A39Z%22%7D%7D%2C%7B%22key%22%3A%22KPUDX7TT%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Han%20et%20al.%22%2C%22parsedDate%22%3A%222026-05-01%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHan%2C%20D.%2C%20Shepherd%2C%20C.%2C%20Benton%2C%20M.%20L.%20%26amp%3B%20Nordman%2C%20J.%20T.%20Nanopore-based%20sequencing%20of%20active%20DNA%20replication%20reveals%20key%20principles%20of%20metazoan%20replication%20dynamics.%20%26lt%3Bi%26gt%3BSci.%20Adv.%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B12%26lt%3B%5C%2Fb%26gt%3B%2C%20eaed2806%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Nanopore-based%20sequencing%20of%20active%20DNA%20replication%20reveals%20key%20principles%20of%20metazoan%20replication%20dynamics%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Dongsheng%22%2C%22lastName%22%3A%22Han%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cole%22%2C%22lastName%22%3A%22Shepherd%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mary%20Lauren%22%2C%22lastName%22%3A%22Benton%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jared%20T.%22%2C%22lastName%22%3A%22Nordman%22%7D%5D%2C%22abstractNote%22%3A%22Balancing%20replication%20fork%20progression%20and%20origin%20usage%20is%20essential%20to%20maintain%20genome%20stability%2C%20but%20measuring%20replication%20fork%20progression%20rates%20and%20origin%20usage%20throughout%20the%20genome%20has%20been%20challenging.%20Here%2C%20we%20use%20nanopore%20sequencing%20combined%20with%20DNAscent%20to%20measure%20replication%20fork%20progression%20together%20with%20origin%20and%20termination%20site%20usage%20with%20single-molecule%20precision%20throughout%20the%20Drosophila%20genome%20with%20nearly%20full%20genome%20coverage.%20We%20find%20that%20replication%20fork%20progression%20rates%20are%20not%20uniform%20throughout%20the%20genome.%20Rather%2C%20fork%20progression%20is%20slowest%20in%20euchromatin%2C%20and%20this%20is%20not%20correlated%20with%20active%20transcription.%20Replication%20origins%20are%20also%20influenced%20by%20chromatin%2C%20but%20the%20exact%20position%20of%20initiation%20is%20highly%20variable%20and%20are%20often%20several%20kilobases%20away%20from%20ORC%20%28origin%20recognition%20complex%29%20binding%20sites.%20Termination%20sites%20lack%20any%20chromatin%20or%20sequence%20motifs%20and%20appear%20nearly%20random%20throughout%20the%20genome.%20By%20measuring%20DNA%20replication%20dynamics%20at%20near%20full%20genome%20coverage%2C%20our%20work%20reveals%20key%20principles%20of%20metazoan%20replication%20dynamics.%20%5Cn%20%20%20%20%20%20%20%20%20%20%2C%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20Using%20nanopore-based%20sequencing%20to%20understand%20metazoan%20replication%20fork%20progression%2C%20initiation%2C%20and%20termination.%22%2C%22date%22%3A%222026-05-01%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1126%5C%2Fsciadv.aed2806%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fwww.science.org%5C%2Fdoi%5C%2F10.1126%5C%2Fsciadv.aed2806%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222375-2548%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A58%3A57Z%22%7D%7D%2C%7B%22key%22%3A%22F2HQABSH%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Ma%20and%20Tuninetti%22%2C%22parsedDate%22%3A%222026-04-29%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMa%2C%20Y.%20%26amp%3B%20Tuninetti%2C%20D.%20On%20Coded%20Caching%20Systems%20with%20Decentralized%20Linear%20Coding%20Placement.%20%26lt%3Bi%26gt%3BarXiv.org%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.27073v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.27073v1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22On%20Coded%20Caching%20Systems%20with%20Decentralized%20Linear%20Coding%20Placement%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yinbin%22%2C%22lastName%22%3A%22Ma%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Daniela%22%2C%22lastName%22%3A%22Tuninetti%22%7D%5D%2C%22abstractNote%22%3A%22Coded%20caching%20is%20a%20technique%20that%20leverages%20locally%20cached%20contents%20at%20the%20end%20users%20to%20reduce%20the%20network%26%23039%3Bs%20peak-time%20communication%20load.%20Coded%20caching%20has%20been%20shown%20to%20achieve%20significant%20performance%20gains%20with%20a%20centralized%20placement%20orchestrated%20by%20the%20server%20and%20is%20thus%20considered%20a%20promising%20technique%20to%20boost%20performance%20in%20future%20networks%20by%20effectively%20trading%20off%20bandwidth%20for%20storage.%20To%20tackle%20issues%20caused%20by%20the%20synchronized%20placement%2C%20previous%20works%20focused%20on%20decentralized%20placement%20and%20found%20the%20exact%20worst-case%20load%20with%20uncoded%20placement.%20In%20this%20paper%2C%20we%20focus%20on%20a%20decentralized%20coded%20caching%20system%20with%20random%20linear%20coding%20placement%2C%20and%20investigate%20the%20fundamental%20limits%20of%20a%20linear%20coding%20placement%20where%20each%20user%20independently%20and%20uniformly%20caches%20random%20linear%20coding%20symbols%20of%20a%20single%20file.%20We%20propose%20achievable%20and%20converse%20bounds%20on%20the%20worst-case%20load%2C%20which%20are%20shown%20to%20meet%20under%20certain%20conditions.%22%2C%22date%22%3A%222026%5C%2F04%5C%2F29%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.27073v1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A58%3A50Z%22%7D%7D%2C%7B%22key%22%3A%22SAQUD3PF%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wu%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-29%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWu%2C%20J.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Eccentricity%20as%20a%20signature%20of%20hierarchical%20subsolar-mass%20mergers%20in%20collapsar%20disks.%20%26lt%3Bi%26gt%3BarXiv.org%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.26912v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.26912v1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22Eccentricity%20as%20a%20signature%20of%20hierarchical%20subsolar-mass%20mergers%20in%20collapsar%20disks%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiaxi%22%2C%22lastName%22%3A%22Wu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Elias%20R.%22%2C%22lastName%22%3A%22Most%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nils%20L.%22%2C%22lastName%22%3A%22Vu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nils%22%2C%22lastName%22%3A%22Deppe%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lawrence%20E.%22%2C%22lastName%22%3A%22Kidder%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kyle%20C.%22%2C%22lastName%22%3A%22Nelli%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%22%2C%22lastName%22%3A%22Throwe%22%7D%5D%2C%22abstractNote%22%3A%22In%20this%20work%2C%20we%20investigate%20gravitational-wave%20signatures%20of%20a%20proposed%20subsolar-mass%20merger%20scenario%20resulting%20from%20fragmentation%20inside%20a%20collapsar%20accretion%20disk.%20This%20scenario%20has%20gained%20recent%20interest%20with%20the%20electromagnetic%20transient%20AT2025ulz%2C%20a%20possible%20superkilonova%20counterpart%20candidate%20to%20the%20sub-threshold%20gravitational%20wave%20event%20S250818k.%20One%20prediction%20of%20fragmentation%20is%20the%20formation%20of%20multiple%20smaller%20neutron-star%20fragments%2C%20some%20of%20which%20might%20merge%20hierarchically.%20Such%20mergers%20are%20expected%20not%20only%20to%20produce%20individual%20electromagnetic%20counterparts%2C%20but%20also%2C%20because%20of%20their%20repeated%20capture%20and%20merger%20dynamics%2C%20to%20impart%20kicks%20to%20the%20system%20and%20thereby%20drive%20orbital%20eccentricity.%20By%20performing%20numerical%20relativity%20simulations%20of%20hierarchical%20compact%20object%20mergers%20modeled%20as%20black%20holes%20in%20a%20disk-like%20geometry%20consistent%20with%20this%20scenario%2C%20we%20demonstrate%20the%20build-up%20of%20potentially%20large%20eccentricity%20for%20the%20final%20merger%2C%20of%20order%20%24e%20%5C%5Csimeq%200.6%24%20initially%2C%20and%20show%20that%2C%20because%20of%20the%20short%20lifetime%20of%20the%20system%2C%20a%20substantial%20part%20of%20this%20eccentricity%20%2C%20up%20to%20%24e%5C%5Csimeq%200.1%24%2C%20can%20survive%20until%20merger%20in%20the%20general%20case.%20As%20a%20result%2C%20future%20detections%20of%20eccentricities%20in%20potential%20subsolar-mass%20gravitational-wave%20candidate%20events%20would%20be%20a%20strong%20indicator%20for%20a%20hierarchical%20formation%20scenario.%22%2C%22date%22%3A%222026%5C%2F04%5C%2F29%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.26912v1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A41%3A21Z%22%7D%7D%2C%7B%22key%22%3A%22DXG3X97J%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wang%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-27%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWang%2C%20B.%2C%20Xu%2C%20G.%2C%20Tang%2C%20Z.%2C%20Zhang%2C%20J.%20%26amp%3B%20Cheng%2C%20Z.%20OmniShotCut%3A%20Holistic%20Relational%20Shot%20Boundary%20Detection%20with%20Shot-Query%20Transformer.%20%26lt%3Bi%26gt%3BarXiv.org%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.24762v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.24762v1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22OmniShotCut%3A%20Holistic%20Relational%20Shot%20Boundary%20Detection%20with%20Shot-Query%20Transformer%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Boyang%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Guangyi%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhipeng%22%2C%22lastName%22%3A%22Tang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiahui%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zezhou%22%2C%22lastName%22%3A%22Cheng%22%7D%5D%2C%22abstractNote%22%3A%22Shot%20Boundary%20Detection%20%28SBD%29%20aims%20to%20automatically%20identify%20shot%20changes%20and%20divide%20a%20video%20into%20coherent%20shots.%20While%20SBD%20was%20widely%20studied%20in%20the%20literature%2C%20existing%20state-of-the-art%20methods%20often%20produce%20non-interpretable%20boundaries%20on%20transitions%2C%20miss%20subtle%20yet%20harmful%20discontinuities%2C%20and%20rely%20on%20noisy%2C%20low-diversity%20annotations%20and%20outdated%20benchmarks.%20To%20alleviate%20these%20limitations%2C%20we%20propose%20OmniShotCut%20to%20formulate%20SBD%20as%20structured%20relational%20prediction%2C%20jointly%20estimating%20shot%20ranges%20with%20intra-shot%20relations%20and%20inter-shot%20relations%2C%20by%20a%20shot%20query-based%20dense%20video%20Transformer.%20To%20avoid%20imprecise%20manual%20labeling%2C%20we%20adopt%20a%20fully%20synthetic%20transition%20synthesis%20pipeline%20that%20automatically%20reproduces%20major%20transition%20families%20with%20precise%20boundaries%20and%20parameterized%20variants.%20We%20also%20introduce%20OmniShotCutBench%2C%20a%20modern%20wide-domain%20benchmark%20enabling%20holistic%20and%20diagnostic%20evaluation.%22%2C%22date%22%3A%222026%5C%2F04%5C%2F27%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.24762v1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A39%3A21Z%22%7D%7D%2C%7B%22key%22%3A%22QF5WF9FD%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Gao%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-24%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BGao%2C%20R.%2C%20Jeon%2C%20Y.%2C%20Roy%2C%20S.%2C%20Ziyadi%2C%20M.%20%26amp%3B%20Chen%2C%20X.%20%26%23x2018%3BAnthony%26%23x2019%3B.%20C-MORAL%3A%20Controllable%20Multi-Objective%20Molecular%20Optimization%20with%20Reinforcement%20Alignment%20for%20LLMs.%20%26lt%3Bi%26gt%3BarXiv.org%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.23061v1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.23061v1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22webpage%22%2C%22title%22%3A%22C-MORAL%3A%20Controllable%20Multi-Objective%20Molecular%20Optimization%20with%20Reinforcement%20Alignment%20for%20LLMs%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rui%22%2C%22lastName%22%3A%22Gao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Youngseung%22%2C%22lastName%22%3A%22Jeon%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Swastik%22%2C%22lastName%22%3A%22Roy%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Morteza%22%2C%22lastName%22%3A%22Ziyadi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiang%20%27Anthony%27%22%2C%22lastName%22%3A%22Chen%22%7D%5D%2C%22abstractNote%22%3A%22Large%20language%20models%20%28LLMs%29%20show%20promise%20for%20molecular%20optimization%2C%20but%20aligning%20them%20with%20selective%20and%20competing%20drug-design%20constraints%20remains%20challenging.%20We%20propose%20C-Moral%2C%20a%20reinforcement%20learning%20post-training%20framework%20for%20controllable%20multi-objective%20molecular%20optimization.%20C-Moral%20combines%20group-based%20relative%20optimization%2C%20property%20score%20alignment%20for%20heterogeneous%20objectives%2C%20and%20continuous%20non-linear%20reward%20aggregation%20to%20improve%20stability%20across%20competing%20properties.%20Experiments%20on%20the%20C-MuMOInstruct%20benchmark%20show%20that%20C-Moral%20consistently%20outperforms%20state-of-the-art%20models%20across%20both%20in-domain%20and%20out-of-domain%20settings%2C%20achieving%20the%20best%20Success%20Optimized%20Rate%20%28SOR%29%20of%2048.9%25%20on%20IND%20tasks%20and%2039.5%25%20on%20OOD%20tasks%2C%20while%20largely%20preserving%20scaffold%20similarity.%20These%20results%20suggest%20that%20RL%20post-training%20is%20an%20effective%20way%20to%20align%20molecular%20language%20models%20with%20continuous%20molecular%20design%20objectives.%20Our%20code%20and%20models%20are%20publicly%20available%20at%20https%3A%5C%2F%5C%2Fgithub.com%5C%2FRwigie%5C%2FC-MORAL.%22%2C%22date%22%3A%222026%5C%2F04%5C%2F24%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.23061v1%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A35%3A12Z%22%7D%7D%2C%7B%22key%22%3A%226G3GEGTB%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Najafi%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-30%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BNajafi%2C%20A.%20R.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Characterization%20and%20Computational%20Engineering%20of%20Structural%20Elements%20Controlling%20Gas%20Permeability%20in%20%26lt%3Bspan%20style%3D%26quot%3Bfont-variant%3Asmall-caps%3B%26quot%3B%26gt%3BPIP2%26lt%3B%5C%2Fspan%26gt%3B%26%23x202F%3B%3B1%20Aquaporins.%20%26lt%3Bi%26gt%3BJ%20Comput%20Chem%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B47%26lt%3B%5C%2Fb%26gt%3B%2C%20e70377%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Characterization%20and%20Computational%20Engineering%20of%20Structural%20Elements%20Controlling%20Gas%20Permeability%20in%20%3Cspan%20style%3D%5C%22font-variant%3Asmall-caps%3B%5C%22%3EPIP2%3C%5C%2Fspan%3E%20%3B1%20Aquaporins%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ahmad%20Raeisi%22%2C%22lastName%22%3A%22Najafi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Paween%22%2C%22lastName%22%3A%22Mahinthichaichan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Fraser%20J.%22%2C%22lastName%22%3A%22Moss%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ardeschir%22%2C%22lastName%22%3A%22Vahedi%5Cu2010Faridi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Walter%20F.%22%2C%22lastName%22%3A%22Boron%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Emad%22%2C%22lastName%22%3A%22Tajkhorshid%22%7D%5D%2C%22abstractNote%22%3A%22ABSTRACT%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20Aquaporins%20%28AQPs%29%20are%20classical%20water%20channels%20that%20also%20conduct%20small%20gas%20molecules%20such%20as%20%20and%20%20across%20the%20membrane.%20The%20hydrophobic%20central%20pore%2C%20located%20at%20the%20fourfold%20symmetry%20axis%20of%20an%20AQP%20tetrameric%20architecture%2C%20has%20been%20proposed%20to%20constitute%20the%20most%20optimal%20pathway%20for%20gas%20transport%2C%20although%20monomeric%20water%20pores%20can%20also%20contribute%20somewhat%20to%20permeation%20of%20less%20hydrophobic%20species.%20Here%2C%20we%20report%20a%20comparative%20molecular%20dynamics%20%28MD%29%20study%20of%20gas%20permeability%20in%20a%20plant%20AQP%20and%20a%20mammalian%20AQP1%2C%20taking%20advantage%20of%20complementary%20computational%20protocols%20including%20flooding%20simulations%2C%20umbrella%20sampling%2C%20and%20implicit%20ligand%20sampling.%20PIP2%3B1%20AQPs%2C%20present%20in%20plants%2C%20are%20experimentally%20reported%20to%20have%20lower%20gas%20permeability%20than%20AQP1%2C%20which%20is%20present%20both%20in%20plants%20and%20animals.%20Using%20the%20spinach%20PIP2%3B1%20%28SoPIP2%3B1%29%20and%20bovine%20AQP1%20%28bAQP1%29%20as%20the%20models%2C%20the%20study%20unravels%20the%20specific%20structural%20features%20controlling%20the%20permeability%20of%20the%20central%20pore%20to%20gases.%20In%20SoPIP2%3B1%2C%20residue%20Trp79%2C%20which%20is%20highly%20conserved%20in%20the%20plant%20PIP2%3B1%20family%20and%20lines%20directly%20the%20central%20pore%2C%20forms%20a%20major%20constriction%20region%20and%20the%20main%20barrier%20against%20gas%20permeation.%20Notably%2C%20the%20occluding%20conformation%20of%20the%20four%20Trp79%20residues%20from%20the%20four%20monomers%20is%20stabilized%20by%20another%20conserved%20residue%2C%20Phe207%20in%20the%20central%20pore.%20Sequence%20and%20structural%20comparisons%20show%20that%20both%20of%20these%20residues%20are%20replaced%20by%20less%20bulky%20residues%20in%20AQP1%2C%20for%20example%2C%20by%20Leu56%20and%20Ala179%2C%20respectively%2C%20in%20bAQP1.%20The%20role%20of%20Phe207%20residues%20in%20hindering%20gas%20permeation%20through%20SoPIP2%3B1%20is%20confirmed%20by%20in%20silico%20alanine%20substitution%2C%20which%20reveals%20its%20effect%20on%20the%20local%20constriction%20produced%20by%20Trp79%20residues.%20Conversely%2C%20by%20mutating%20Leu56%20to%20tryptophan%20and%20Ala179%20to%20phenylalanine%20in%20bAQP1%2C%20we%20engineer%20the%20protein%20to%20a%20less%20permeable%20gas%20channel.%22%2C%22date%22%3A%222026-04-30%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1002%5C%2Fjcc.70377%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fonlinelibrary.wiley.com%5C%2Fdoi%5C%2F10.1002%5C%2Fjcc.70377%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%220192-8651%2C%201096-987X%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A24%3A25Z%22%7D%7D%2C%7B%22key%22%3A%22HA2TFBS5%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Kalyanasundaram%20and%20Chaspari%22%2C%22parsedDate%22%3A%222026-05-03%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BKalyanasundaram%2C%20S.%20%26amp%3B%20Chaspari%2C%20T.%20Conditional%20Diffusion%20Models%20for%20Mental%20Health-Preserving%20Voice%20Conversion.%20in%20%26lt%3Bi%26gt%3BICASSP%202026%20-%202026%20IEEE%20International%20Conference%20on%20Acoustics%2C%20Speech%20and%20Signal%20Processing%20%28ICASSP%29%26lt%3B%5C%2Fi%26gt%3B%2017982%26%23x2013%3B17986%20%28IEEE%2C%20Barcelona%2C%20Spain%2C%202026%29.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1109%5C%2FICASSP55912.2026.11464847%26%23039%3B%26gt%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1109%5C%2FICASSP55912.2026.11464847%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Conditional%20Diffusion%20Models%20for%20Mental%20Health-Preserving%20Voice%20Conversion%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Siddharth%22%2C%22lastName%22%3A%22Kalyanasundaram%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Theodora%22%2C%22lastName%22%3A%22Chaspari%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22proceedingsTitle%22%3A%22ICASSP%202026%20-%202026%20IEEE%20International%20Conference%20on%20Acoustics%2C%20Speech%20and%20Signal%20Processing%20%28ICASSP%29%22%2C%22conferenceName%22%3A%22ICASSP%202026%20-%202026%20IEEE%20International%20Conference%20on%20Acoustics%2C%20Speech%20and%20Signal%20Processing%20%28ICASSP%29%22%2C%22date%22%3A%222026-5-3%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%2210.1109%5C%2FICASSP55912.2026.11464847%22%2C%22ISBN%22%3A%229798331567019%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fieeexplore.ieee.org%5C%2Fdocument%5C%2F11464847%5C%2F%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T16%3A15%3A06Z%22%7D%7D%2C%7B%22key%22%3A%22W26T5G3H%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Howes%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-17%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHowes%2C%20W.%2C%20Ahmed%2C%20F.%2C%20Kobayashi%2C%20K.%2C%20Chakraborty%2C%20S.%20%26amp%3B%20Alam%2C%20S.%20B.%20Neuroscience%20Inspired%20Graph%20Operators%20Towards%20Edge-Deployable%20Virtual%20Sensing%20for%20Irregular%20Geometries.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2604.16722%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2604.16722%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Neuroscience%20Inspired%20Graph%20Operators%20Towards%20Edge-Deployable%20Virtual%20Sensing%20for%20Irregular%20Geometries%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%22%2C%22lastName%22%3A%22Howes%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Farid%22%2C%22lastName%22%3A%22Ahmed%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kazuma%22%2C%22lastName%22%3A%22Kobayashi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Souvik%22%2C%22lastName%22%3A%22Chakraborty%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Syed%20Bahauddin%22%2C%22lastName%22%3A%22Alam%22%7D%5D%2C%22abstractNote%22%3A%22Predicting%20full-field%20physics%20through%20the%20real-time%20virtual%20sensing%20of%20engineering%20systems%20can%20enhance%20limited%20physical%20sensors%20but%20often%20requires%20sparse-to-dense%20reconstruction%2C%20complex%20multiphysics%2C%20and%20highly%20irregular%20geometries%20as%20well%20as%20strict%20latency%20and%20energy%20constraints%20for%20edge-deployability.%20Neural%20operators%20have%20been%20presented%20as%20a%20potential%20candidate%20for%20such%20applications%20but%20few%20architectures%20exist%20that%20explicitly%20address%20power%20consumption.%20Spiking%20neuron%20integration%20can%20provide%20a%20potential%20solution%20when%20integrated%20on%20neuromorphic%20hardware%20but%20the%20current%20existing%20neuron%20models%20result%20in%20severe%20performance%20degradation%20towards%20regression-based%20virtual%20sensing.%20To%20address%20the%20performance%20concerns%20and%20edge-constraints%2C%20we%20present%20the%20Variable%20Spiking%20Graph%20Neural%20Operator%20%28VS-GNO%29%20which%20integrates%20a%20sophisticated%20spectral-spatial%20convolutional%20analysis%20and%20a%20previously%20developed%20Variable%20Spiking%20Neuron%20%28VSN%29%20and%20energy-error%20balance%20loss%20function.%20With%20a%20non-spiking%20%24L_2%24%20error%20baseline%20of%20%240.4%5C%5C%25%24%2C%20VS-GNO%20can%20provide%20a%20reconstruction%20error%20of%20%240.71%5C%5C%25%24%20with%20%2415%5C%5C%25%24%20average%20spiking%20in%20its%20spectral-only%20form%20and%20%241.04%5C%5C%25%24%20with%20%2424.5%5C%5C%25%24%20spiking%20in%20its%20entire%20form.%20These%20results%20position%20VS-GNO%20as%20a%20promising%20step%20towards%20energy-efficient%2C%20edge-deployable%20neural%20operators%20for%20real-time%20sparse-to-dense%20virtual%20sensing%20in%20complex%2C%20highly%20irregular%20engineering%20environments.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22arXiv%3A2604.16722%22%2C%22date%22%3A%222026-04-17%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2604.16722%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22http%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.16722%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-11T15%3A51%3A26Z%22%7D%7D%2C%7B%22key%22%3A%22M4R6X7FN%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Dong%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BDong%2C%20P.%2C%20Swerdlow%2C%20A.%2C%20Sadigh%2C%20D.%20%26amp%3B%20Finn%2C%20C.%20FASTER%3A%20Value-Guided%20Sampling%20for%20Fast%20RL.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.19730%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.19730%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22FASTER%3A%20Value-Guided%20Sampling%20for%20Fast%20RL%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Perry%22%2C%22lastName%22%3A%22Dong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alexander%22%2C%22lastName%22%3A%22Swerdlow%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Dorsa%22%2C%22lastName%22%3A%22Sadigh%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chelsea%22%2C%22lastName%22%3A%22Finn%22%7D%5D%2C%22abstractNote%22%3A%22Some%20of%20the%20most%20performant%20reinforcement%20learning%20algorithms%20today%20can%20be%20prohibitively%20expensive%20as%20they%20use%20test-time%20scaling%20methods%20such%20as%20sampling%20multiple%20action%20candidates%20and%20selecting%20the%20best%20one.%20In%20this%20work%2C%20we%20propose%20FASTER%2C%20a%20method%20for%20getting%20the%20benefits%20of%20sampling-based%20test-time%20scaling%20of%20diffusion-based%20policies%20without%20the%20computational%20cost%20by%20tracing%20the%20performance%20gain%20of%20action%20samples%20back%20to%20earlier%20in%20the%20denoising%20process.%20Our%20key%20insight%20is%20that%20we%20can%20model%20the%20denoising%20of%20multiple%20action%20candidates%20and%20selecting%20the%20best%20one%20as%20a%20Markov%20Decision%20Process%20%28MDP%29%20where%20the%20goal%20is%20to%20progressively%20filter%20action%20candidates%20before%20denoising%20is%20complete.%20With%20this%20MDP%2C%20we%20can%20learn%20a%20policy%20and%20value%20function%20in%20the%20denoising%20space%20that%20predicts%20the%20downstream%20value%20of%20action%20candidates%20in%20the%20denoising%20process%20and%20filters%20them%20while%20maximizing%20returns.%20The%20result%20is%20a%20method%20that%20is%20lightweight%20and%20can%20be%20plugged%20into%20existing%20generative%20RL%20algorithms.%20Across%20challenging%20long-horizon%20manipulation%20tasks%20in%20online%20and%20batch-online%20RL%2C%20FASTER%20consistently%20improves%20the%20underlying%20policies%20and%20achieves%20the%20best%20overall%20performance%20among%20the%20compared%20methods.%20Applied%20to%20a%20pretrained%20VLA%2C%20FASTER%20achieves%20the%20same%20performance%20while%20substantially%20reducing%20training%20and%20inference%20compute%20requirements.%20Code%20is%20available%20at%20https%3A%5C%2F%5C%2Fgithub.com%5C%2Falexanderswerdlow%5C%2Ffaster%20.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.19730%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.19730%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T22%3A28%3A40Z%22%7D%7D%2C%7B%22key%22%3A%222CQG3SL7%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Kaushal%20and%20Penmatsa%22%2C%22parsedDate%22%3A%222026-03-19%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BKaushal%2C%20N.%20%26amp%3B%20Penmatsa%2C%20A.%20M.%20FragBERTa%3A%20Exploring%20Fragment-based%20Molecular%20Representation%20Learning%20with%20SAFE.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.26434%5C%2Fchemrxiv.15000476%5C%2Fv2%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.26434%5C%2Fchemrxiv.15000476%5C%2Fv2%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22FragBERTa%3A%20Exploring%20Fragment-based%20Molecular%20Representation%20Learning%20with%20SAFE%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Neerav%22%2C%22lastName%22%3A%22Kaushal%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ajay%20Mnv%22%2C%22lastName%22%3A%22Penmatsa%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026-03-19%22%2C%22DOI%22%3A%2210.26434%5C%2Fchemrxiv.15000476%5C%2Fv2%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fchemrxiv.org%5C%2Fdoi%5C%2Ffull%5C%2F10.26434%5C%2Fchemrxiv.15000476%5C%2Fv2%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T22%3A26%3A56Z%22%7D%7D%2C%7B%22key%22%3A%22F46FPRSW%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Bouckley%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BBouckley%2C%20E.%20R.%2C%20Lewin%2C%20S.%20F.%20%26amp%3B%20Lefauve%2C%20A.%20Early%20onset%20of%20secondary%20shear%20instability%20in%20Kelvin-Helmholtz%20braids%20at%20high%20Reynolds%20number.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.16173%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.16173%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Early%20onset%20of%20secondary%20shear%20instability%20in%20Kelvin-Helmholtz%20braids%20at%20high%20Reynolds%20number%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Emma%20R.%22%2C%22lastName%22%3A%22Bouckley%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sam%20F.%22%2C%22lastName%22%3A%22Lewin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Adrien%22%2C%22lastName%22%3A%22Lefauve%22%7D%5D%2C%22abstractNote%22%3A%22We%20study%20the%20onset%20of%20two-dimensional%20secondary%20shear%20instability%20%28SSI%29%20in%20the%20braid%20regions%20connecting%20primary%20Kelvin-Helmholtz%20billows%20in%20stratified%20shear%20flows.%20While%20strain%20induced%20by%20the%20billows%20stabilises%20the%20braids%2C%20it%20also%20compresses%20their%20tilted%20isopycnals%2C%20enhancing%20baroclinic%20shear%20that%20enables%20rapid%20perturbation%20growth.%20By%20modifying%20the%20classical%20analysis%20of%20Corcos%20%26amp%3B%20Sherman%20%28J.%20Fluid%20Mech.%2073%2C%20241-264%2C%201976%29%20in%20braid-aligned%20coordinates%20and%20adding%20an%20additional%20stability%20criterion%20based%20on%20the%20ratio%20of%20strain%20rate%20to%20shear%2C%20we%20develop%20an%20inviscid%2C%20time-dependent%20model%20for%20the%20braid%20and%20the%20onset%20of%20SSI.%20We%20show%20that%20the%20criterion%20for%20instability%20can%20be%20achieved%20significantly%20earlier%20than%20the%20saturation%20of%20the%20primary%20billow%20at%20sufficiently%20high%20initial%20Richardson%20number%20Ri%2C%20as%20increased%20stratification%20slows%20billow%20growth%20while%20accelerating%20baroclinic%20shear%20production%20in%20the%20braid.%20Two-dimensional%20direct%20numerical%20simulations%20up%20to%20Reynolds%20numbers%20Re%3D10%5E7%20quantify%20the%20role%20of%20viscosity.%20At%20high%20Re%2C%20we%20find%20that%20SSI%20indeed%20develops%20early%20in%20the%20braid%2C%20as%20predicted%20by%20the%20inviscid%20model%2C%20while%20the%20primary%20billow%20is%20still%20growing%20and%20before%20viscosity%20slows%20braid%20thinning.%20These%20results%20provide%20a%20mechanistic%20explanation%20for%20field%20observations%20of%20braid-dominated%20mixing%20and%20suggest%20that%2C%20at%20geophysically%20relevant%20Ri%20and%20Re%2C%20SSI%20can%20control%20the%20three-dimensional%20turbulent%20transition%20and%20ensuing%20diapycnal%20mixing%20by%20preceding%20and%20pre-empting%20both%20vortex%20pairing%20instabilities%20and%20secondary%20convective%20instabilities%20in%20the%20billow%20core.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.16173%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.16173%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T22%3A25%3A13Z%22%7D%7D%2C%7B%22key%22%3A%2237WFYKUA%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Bush%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BBush%2C%20M.%2C%20Palmer%2C%20J.%20C.%20%26amp%3B%20Howard%2C%20M.%20P.%20Simulating%20hydrodynamic%20interactions%20in%20colloidal%20suspensions%20using%20multiparticle%20collision%20dynamics%20with%20rigid-body%20constraints.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.14307%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.14307%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Simulating%20hydrodynamic%20interactions%20in%20colloidal%20suspensions%20using%20multiparticle%20collision%20dynamics%20with%20rigid-body%20constraints%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michaela%22%2C%22lastName%22%3A%22Bush%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jeremy%20C.%22%2C%22lastName%22%3A%22Palmer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%20P.%22%2C%22lastName%22%3A%22Howard%22%7D%5D%2C%22abstractNote%22%3A%22We%20develop%20a%20method%20for%20simulating%20colloidal%20suspensions%20using%20multiparticle%20collision%20dynamics%20%28MPCD%29%20with%20a%20discrete%20particle%20model%20represented%20as%20a%20rigid%20body.%20The%20key%20steps%20for%20incorporating%20the%20rigid-body%20constraints%20are%20to%20thermalize%20the%20velocities%20of%20the%20discrete%20sites%20before%20they%20participate%20in%20the%20MPCD%20collision%20step%2C%20then%20transfer%20momentum%20from%20the%20sites%20to%20the%20rigid%20body.%20We%20demonstrate%20that%20the%20rigid-body%20model%20produces%20the%20expected%20statistics%20for%20a%20single%20spherical%20particle%20and%20the%20same%20transport%20properties%20for%20a%20hard-sphere%20colloidal%20suspension%20as%20an%20equivalent%20model%20using%20harmonic%20bonds%20to%20maintain%20the%20site%20geometry.%20Importantly%2C%20the%20rigid-body%20model%20has%20less%20computational%20overhead%20and%20permits%20a%20larger%20simulation%20timestep%20than%20the%20harmonic-bond%20model%2C%20leading%20to%20a%20nearly%20order%20of%20magnitude%20speedup%20in%20benchmark%20simulations%20of%20hard-sphere%20colloidal%20suspensions.%20Our%20method%20is%20compatible%20with%20arbitrary%20discretization%2C%20so%20it%20enables%20more%20efficient%20MPCD%20simulations%20of%20suspensions%20of%20colloidal%20particles%20with%20complex%20shapes.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.14307%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.14307%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T22%3A16%3A13Z%22%7D%7D%2C%7B%22key%22%3A%22FVW3748J%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Roy%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BRoy%2C%20S.%2C%20Chakraborty%2C%20S.%20%26amp%3B%20Alam%2C%20S.%20B.%20Beyond%20Uniform%20Sampling%3A%20Synergistic%20Active%20Learning%20and%20Input%20Denoising%20for%20Robust%20Neural%20Operators.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.13316%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.13316%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Beyond%20Uniform%20Sampling%3A%20Synergistic%20Active%20Learning%20and%20Input%20Denoising%20for%20Robust%20Neural%20Operators%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Samrendra%22%2C%22lastName%22%3A%22Roy%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Souvik%22%2C%22lastName%22%3A%22Chakraborty%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Syed%20Bahauddin%22%2C%22lastName%22%3A%22Alam%22%7D%5D%2C%22abstractNote%22%3A%22Neural%20operators%20have%20emerged%20as%20fast%20surrogate%20models%20for%20physics%20simulations%2C%20yet%20they%20remain%20acutely%20vulnerable%20to%20adversarial%20perturbations%2C%20a%20critical%20liability%20for%20safety-critical%20digital%20twin%20deployments.%20We%20present%20a%20synergistic%20defense%20that%20combines%20active%20learning-based%20data%20generation%20with%20an%20input%20denoising%20architecture.%20The%20active%20learning%20component%20adaptively%20probes%20model%20weaknesses%20using%20differential%20evolution%20attacks%2C%20then%20generates%20targeted%20training%20data%20at%20discovered%20vulnerability%20locations%20while%20an%20adaptive%20smooth-ratio%20safeguard%20preserves%20baseline%20accuracy.%20The%20input%20denoising%20component%20augments%20the%20operator%20architecture%20with%20a%20learnable%20bottleneck%20that%20filters%20adversarial%20noise%20while%20retaining%20physics-relevant%20features.%20On%20the%20viscous%20Burgers%26%23039%3B%20equation%20benchmark%2C%20the%20combined%20approach%20achieves%20a%202.04%25%20combined%20error%20%281.21%25%20baseline%20%2B%200.83%25%20robustness%29%2C%20representing%20an%2087%25%20reduction%20relative%20to%20standard%20training%20%2815.42%25%20combined%29%20and%20outperforming%20both%20active%20learning%20alone%20%283.42%25%29%20and%20input%20denoising%20alone%20%285.22%25%29.%20More%20broadly%2C%20our%20results%2C%20combined%20with%20cross-architecture%20vulnerability%20analysis%20from%20prior%20work%2C%20suggest%20that%20optimal%20training%20data%20for%20neural%20operators%20is%20architecture-dependent%3A%20because%20different%20architectures%20concentrate%20sensitivity%20in%20distinct%20input%20subspaces%2C%20uniform%20sampling%20cannot%20adequately%20cover%20the%20vulnerability%20landscape%20of%20all%20models.%20These%20findings%20have%20potential%20implications%20for%20the%20deployment%20of%20neural%20operators%20in%20safety-critical%20energy%20systems%20including%20nuclear%20reactor%20monitoring.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.13316%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.13316%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T22%3A04%3A36Z%22%7D%7D%2C%7B%22key%22%3A%22EB7K8SYK%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Motta%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMotta%2C%20P.%20N.%2C%20Neto%2C%20M.%20R.%2C%20Prather%2C%20C.%20%26amp%3B%20C%26%23xE1%3Brdenas-Avenda%26%23xF1%3Bo%2C%20A.%20Sensitivities%20of%20Black%20Hole%20Images%20from%20GRMHD%20Simulations.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.11869%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.11869%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Sensitivities%20of%20Black%20Hole%20Images%20from%20GRMHD%20Simulations%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Pedro%20Naethe%22%2C%22lastName%22%3A%22Motta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22M%5Cu00e1rio%20Raia%22%2C%22lastName%22%3A%22Neto%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cora%22%2C%22lastName%22%3A%22Prather%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alejandro%22%2C%22lastName%22%3A%22C%5Cu00e1rdenas-Avenda%5Cu00f1o%22%7D%5D%2C%22abstractNote%22%3A%22The%20advent%20of%20high-fidelity%20imaging%20of%20supermassive%20black%20holes%20calls%20for%20efficient%20and%20robust%20data-analysis%20methods.%20In%20this%20work%2C%20we%20use%20%24%5C%5Ctexttt%7BJipole%7D%24%2C%20a%20differentiable%2C%20%24%5C%5Ctexttt%7Bipole%7D%24-based%20radiative%20transfer%20code%2C%20to%20enable%20gradient-based%20analyses%20of%20images%20generated%20from%20state-of-the-art%20general%20relativistic%20magnetohydrodynamic%20%28GRMHD%29%20simulations.%20We%20compute%20image%20sensitivities%2C%20i.e.%2C%20pixel-wise%20derivatives%20of%20the%20intensity%20with%20respect%20to%20model%20parameters%2C%20which%20form%20the%20Jacobian%20of%20the%20forward%20model%20and%20define%20a%20local%20map%20from%20parameter%20space%20to%20image%20space.%20Using%20these%20sensitivities%20in%20a%20mock%20data%20analysis%2C%20we%20find%20that%20GRMHD-based%20images%20generate%20a%20structured%20error%20landscape%20for%20parameter%20fitting%2C%20with%20anisotropies%20and%20local%20minima%2C%20making%20parameter%20exploration%20nontrivial%20but%20still%20tractable%20when%20guided%20by%20gradient%20information.%20We%20characterize%20this%20landscape%20through%20the%20Jacobian%20and%20assess%20the%20feasibility%20of%20gradient-based%20recovery%20under%20idealized%2C%20blurred%2C%20and%20noisy%20conditions.%20Our%20results%20show%20that%20automatic%20differentiation-computed%20image%20gradients%20can%20guide%20parameter%20exploration%20effectively%20even%20in%20the%20presence%20of%20noise.%20These%20findings%20establish%20a%20basis%20for%20efficient%2C%20high-precision%20model--data%20comparisons%20in%20black%20hole%20imaging%20and%20motivate%20the%20integration%20of%20these%20sensitivities%20into%20advanced%20inference%20frameworks.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.11869%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.11869%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T22%3A00%3A11Z%22%7D%7D%2C%7B%22key%22%3A%229N3PBSAT%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Shivaiah%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-14%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BShivaiah%2C%20K.-K.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20A%20Conserved%20Sequence%20Insert%20of%20Fibrillins%20is%20Involved%20in%20Plastoglobule%20Association%20and%20Lipid%20Binding.%20%26lt%3Bi%26gt%3BPlant%20and%20Cell%20Physiology%26lt%3B%5C%2Fi%26gt%3B%20pcag047%20%282026%29%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1093%5C%2Fpcp%5C%2Fpcag047%26%23039%3B%26gt%3Bhttp%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1093%5C%2Fpcp%5C%2Fpcag047%26lt%3B%5C%2Fa%26gt%3B.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22A%20Conserved%20Sequence%20Insert%20of%20Fibrillins%20is%20Involved%20in%20Plastoglobule%20Association%20and%20Lipid%20Binding%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kiran-Kumar%22%2C%22lastName%22%3A%22Shivaiah%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Roselane%22%2C%22lastName%22%3A%22Kithan-Lundquist%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22April%20K%22%2C%22lastName%22%3A%22Kaneshiro%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Duncan%20M%22%2C%22lastName%22%3A%22Boren%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Febri%20A%22%2C%22lastName%22%3A%22Susanto%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andres%22%2C%22lastName%22%3A%22Herrera-Tequia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Josh%20V%22%2C%22lastName%22%3A%22Vermaas%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Peter%20Knut%22%2C%22lastName%22%3A%22Lundquist%22%7D%5D%2C%22abstractNote%22%3A%22ABSTRACT%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20Plastoglobule%20lipid%20droplets%20of%20chloroplasts%20serve%20complex%20roles%20affecting%20plant%20development%2C%20stress%20tolerance%20and%20photosynthesis.%20They%20harbor%20a%20set%20of%20approximately%2040%20proteins%20that%20collectively%20dictate%20plastoglobule%20functions.%20Due%20to%20the%20monolayer%20structure%20of%20plastoglobules%2C%20which%20encompass%20a%20neutral%20lipid%20core%2C%20these%20proteins%20must%20associate%20monotopically%20on%20the%20plastoglobule%20surface.%20However%2C%20targeting%20determinants%20have%20not%20been%20identified%20for%20plastoglobule%20proteins%2C%20and%20the%20protein-membrane%20interaction%20mechanisms%20that%20establish%20the%20plastoglobule%20proteome%20remain%20unclear.%20Here%2C%20we%20demonstrate%20that%20most%20plastoglobule-localized%20Fibrillins%20harbor%20an%20extra%20sequence%20insert%20with%20predicted%20amphipathic%20helical%20structure%2C%20placing%20it%20at%20the%20lip%20of%20the%20%5Cu03b2-barrel.%20Deletions%20of%20the%20insert%20in%20two%20plastoglobule-targeted%20FBNs%20disrupt%20their%20localization%2C%20indicating%20that%20they%20are%20necessary%20for%20effective%20plastoglobule%20association.%20Molecular%20dynamics%20simulations%20support%20the%20specific%20interaction%20of%20the%20amphipathic%20helical%20insert%20of%20AtFBN1a%20with%20membranes%20rich%20in%20lipid%20packing%20defects%20which%20are%20expected%20to%20be%20especially%20prevalent%20on%20the%20tightly%20curved%20surface%20of%20plastoglobules.%20Proteomic%20analyses%20indicate%20AtFBN1a%20influences%20the%20plastoglobule%20proteome%20through%20outcompeting%20and%20recruiting%20specific%20proteins.%20We%20also%20demonstrate%20that%20the%20plastoglobule-localized%20FBNs%2C%20AtFBN1a%20and%20AtFBN7a%2C%20bind%20unsaturated%20fatty%20acids%2C%20particularly%20C18%3A1%2C%20and%20that%20elimination%20of%20the%20amphipathic%20helical%20inserts%20suppress%20fatty%20acid%20binding%20in%20AtFBN1a%2C%20but%20promotes%20fatty%20acid%20binding%20in%20AtFBN7a.%20Predicted%20amphipathic%20helices%20can%20be%20identified%20on%20two-thirds%20of%20plastoglobule%20proteins%2C%20indicating%20the%20use%20of%20amphipathic%20helices%20may%20be%20a%20general%20mechanism%20by%20which%20proteins%20selectively%20associate%20with%20plastoglobules.%22%2C%22date%22%3A%222026-04-14%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1093%5C%2Fpcp%5C%2Fpcag047%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Facademic.oup.com%5C%2Fpcp%5C%2Fadvance-article%5C%2Fdoi%5C%2F10.1093%5C%2Fpcp%5C%2Fpcag047%5C%2F8653874%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%220032-0781%2C%201471-9053%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T21%3A51%3A21Z%22%7D%7D%2C%7B%22key%22%3A%22F9RXIV3K%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Naethe%20Motta%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-20%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BNaethe%20Motta%2C%20P.%2C%20Nemmen%2C%20R.%20%26amp%3B%20Joshi%2C%20A.%20V.%20GPUmonty%26%23x202F%3B%3A%20A%20GPU-accelerated%20Relativistic%20Monte%20Carlo%20Radiative%20Transfer%20Code.%20%26lt%3Bi%26gt%3BApJ%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B1001%26lt%3B%5C%2Fb%26gt%3B%2C%20148%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22GPUmonty%20%3A%20A%20GPU-accelerated%20Relativistic%20Monte%20Carlo%20Radiative%20Transfer%20Code%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Pedro%22%2C%22lastName%22%3A%22Naethe%20Motta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rodrigo%22%2C%22lastName%22%3A%22Nemmen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Abhishek%20V.%22%2C%22lastName%22%3A%22Joshi%22%7D%5D%2C%22abstractNote%22%3A%22Abstract%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20We%20introduce%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20GPUmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%2C%20a%20CUDA%5C%2FC-based%20Monte%20Carlo%20radiative%20transfer%20code%20accelerated%20using%20graphics%20processing%20units%20%28GPUs%29.%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20GPUmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20derives%20from%20the%20CPU-based%20code%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20grmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20and%20offloads%20the%20most%20computationally%20expensive%20stages%20of%20the%20calculation%5Cu2014superphoton%20generation%2C%20sampling%2C%20tracking%2C%20and%20scattering%5Cu2014to%20the%20GPU.%20Whereas%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20grmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20handles%20photons%20sequentially%2C%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20GPUmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20processes%20large%20numbers%20of%20superphotons%20concurrently%2C%20leveraging%20the%20single-instruction%2C%20multiple-thread%20execution%20model%20of%20modern%20GPUs.%20Benchmarks%20demonstrate%20a%20speedup%20of%20about%2012%5Cu00d7%20relative%20to%20the%20original%20CPU%20implementation%20on%20a%20single%20GPU%2C%20with%20runtime%20limited%20primarily%20by%20register%20pressure%20rather%20than%20compute%20or%20memory%20bandwidth%20saturation.%20We%20validate%20the%20implementation%20through%20analytic%20tests%20for%20a%20optically%20thin%20synchrotron%20sphere%2C%20as%20well%20as%20comparisons%20with%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20igrmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20for%20scattering%20synchrotron%20sphere%20and%20general%20relativistic%20magnetohydrodynamic%20simulation%20data.%20Relative%20errors%20remain%20below%20a%20percent%20level%2C%20and%20convergence%20is%20consistent%20with%20the%20expected%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20N%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20s%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cu2212%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%201%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5C%2F%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%202%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20Monte%20Carlo%20scaling.%20By%20significantly%20reducing%20computational%20costs%2C%20GPUmonty%20enables%20the%20extensive%20parameter%20space%20surveys%20and%20faster%20spectra%20modeling%20required%20to%20interpret%20horizon-scale%20observations%20of%20supermassive%20black%20holes.%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20GPUmonty%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20%20%20is%20publicly%20available%20under%20the%20GNU%20%28GNU%26%23039%3Bs%20Not%20Unix%29%20General%20Public%20License.%22%2C%22date%22%3A%222026-04-20%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.3847%5C%2F1538-4357%5C%2Fae5631%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fiopscience.iop.org%5C%2Farticle%5C%2F10.3847%5C%2F1538-4357%5C%2Fae5631%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%220004-637X%2C%201538-4357%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T21%3A43%3A14Z%22%7D%7D%2C%7B%22key%22%3A%22YFNMQZI4%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Gu%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BGu%2C%20L.%2C%20Huang%2C%20H.-J.%2C%20Qadri%2C%20M.%2C%20Kaess%2C%20M.%20%26amp%3B%20Yuan%2C%20W.%20TouchAnything%3A%20Diffusion-Guided%203D%20Reconstruction%20from%20Sparse%20Robot%20Touches.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.08945%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.08945%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22TouchAnything%3A%20Diffusion-Guided%203D%20Reconstruction%20from%20Sparse%20Robot%20Touches%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Langzhe%22%2C%22lastName%22%3A%22Gu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Hung-Jui%22%2C%22lastName%22%3A%22Huang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mohamad%22%2C%22lastName%22%3A%22Qadri%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%22%2C%22lastName%22%3A%22Kaess%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Wenzhen%22%2C%22lastName%22%3A%22Yuan%22%7D%5D%2C%22abstractNote%22%3A%22Accurate%20object%20geometry%20estimation%20is%20essential%20for%20many%20downstream%20tasks%2C%20including%20robotic%20manipulation%20and%20physical%20interaction.%20Although%20vision%20is%20the%20dominant%20modality%20for%20shape%20perception%2C%20it%20becomes%20unreliable%20under%20occlusions%20or%20challenging%20lighting%20conditions.%20In%20such%20scenarios%2C%20tactile%20sensing%20provides%20direct%20geometric%20information%20through%20physical%20contact.%20However%2C%20reconstructing%20global%203D%20geometry%20from%20sparse%20local%20touches%20alone%20is%20fundamentally%20underconstrained.%20We%20present%20TouchAnything%2C%20a%20framework%20that%20leverages%20a%20pretrained%20large-scale%202D%20vision%20diffusion%20model%20as%20a%20semantic%20and%20geometric%20prior%20for%203D%20reconstruction%20from%20sparse%20tactile%20measurements.%20Unlike%20prior%20work%20that%20trains%20category-specific%20reconstruction%20networks%20or%20learns%20diffusion%20models%20directly%20from%20tactile%20data%2C%20we%20transfer%20the%20geometric%20knowledge%20encoded%20in%20pretrained%20visual%20diffusion%20models%20to%20the%20tactile%20domain.%20Given%20sparse%20contact%20constraints%20and%20a%20coarse%20class-level%20description%20of%20the%20object%2C%20we%20formulate%20reconstruction%20as%20an%20optimization%20problem%20that%20enforces%20tactile%20consistency%20while%20guiding%20solutions%20toward%20shapes%20consistent%20with%20the%20diffusion%20prior.%20Our%20method%20reconstructs%20accurate%20geometries%20from%20only%20a%20few%20touches%2C%20outperforms%20existing%20baselines%2C%20and%20enables%20open-world%203D%20reconstruction%20of%20previously%20unseen%20object%20instances.%20Our%20project%20page%20is%20https%3A%5C%2F%5C%2Fgrange007.github.io%5C%2Ftouchanything%20.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.08945%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.08945%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T21%3A04%3A58Z%22%7D%7D%2C%7B%22key%22%3A%22SLCYWF98%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Liu%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLiu%2C%20E.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20What%20do%20Language%20Models%20Learn%20and%20When%3F%20The%20Implicit%20Curriculum%20Hypothesis.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.08510%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.08510%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22What%20do%20Language%20Models%20Learn%20and%20When%3F%20The%20Implicit%20Curriculum%20Hypothesis%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Emmy%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kaiser%22%2C%22lastName%22%3A%22Sun%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Millicent%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Isabelle%22%2C%22lastName%22%3A%22Lee%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Lindia%22%2C%22lastName%22%3A%22Tjuatja%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jen-tse%22%2C%22lastName%22%3A%22Huang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Graham%22%2C%22lastName%22%3A%22Neubig%22%7D%5D%2C%22abstractNote%22%3A%22Large%20language%20models%20%28LLMs%29%20can%20perform%20remarkably%20complex%20tasks%2C%20yet%20the%20fine-grained%20details%20of%20how%20these%20capabilities%20emerge%20during%20pretraining%20remain%20poorly%20understood.%20Scaling%20laws%20on%20validation%20loss%20tell%20us%20how%20much%20a%20model%20improves%20with%20additional%20compute%2C%20but%20not%20what%20skills%20it%20acquires%20in%20which%20order.%20To%20remedy%20this%2C%20we%20propose%20the%20Implicit%20Curriculum%20Hypothesis%3A%20pretraining%20follows%20a%20compositional%20and%20predictable%20curriculum%20across%20models%20and%20data%20mixtures.%20We%20test%20this%20by%20designing%20a%20suite%20of%20simple%2C%20composable%20tasks%20spanning%20retrieval%2C%20morphological%20transformations%2C%20coreference%2C%20logical%20reasoning%2C%20and%20mathematics.%20Using%20these%20tasks%2C%20we%20track%20emergence%20points%20across%20four%20model%20families%20spanning%20sizes%20from%20410M-13B%20parameters.%20We%20find%20that%20emergence%20orderings%20of%20when%20models%20reach%20fixed%20accuracy%20thresholds%20are%20strikingly%20consistent%20%28%24%5Cu03c1%3D%20.81%24%20across%2045%20model%20pairs%29%2C%20and%20that%20composite%20tasks%20most%20often%20emerge%20after%20their%20component%20tasks.%20Furthermore%2C%20we%20find%20that%20this%20structure%20is%20encoded%20in%20model%20representations%3A%20tasks%20with%20similar%20function%20vector%20representations%20also%20tend%20to%20follow%20similar%20trajectories%20in%20training.%20By%20using%20the%20space%20of%20representations%20derived%20from%20our%20task%20set%2C%20we%20can%20effectively%20predict%20the%20training%20trajectories%20of%20simple%20held-out%20compositional%20tasks%20throughout%20the%20course%20of%20pretraining%20%28%24R%5E2%20%3D%20.68%24-%24.84%24%20across%20models%29%20without%20previously%20evaluating%20them.%20Together%2C%20these%20results%20suggest%20that%20pretraining%20is%20more%20structured%20than%20loss%20curves%20reveal%3A%20skills%20emerge%20in%20a%20compositional%20order%20that%20is%20consistent%20across%20models%20and%20readable%20from%20their%20internals.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.08510%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.08510%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T20%3A50%3A06Z%22%7D%7D%2C%7B%22key%22%3A%22WTPLJ2N2%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Jayakody%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BJayakody%2C%20S.%2C%20Zhao%2C%20Y.%2C%20Nehate%2C%20C.%20D.%20%26amp%3B%20Wang%2C%20J.%20GhostServe%3A%20A%20Lightweight%20Checkpointing%20System%20in%20the%20Shadow%20for%20Fault-Tolerant%20LLM%20Serving.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2605.00831%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2605.00831%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22GhostServe%3A%20A%20Lightweight%20Checkpointing%20System%20in%20the%20Shadow%20for%20Fault-Tolerant%20LLM%20Serving%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shakya%22%2C%22lastName%22%3A%22Jayakody%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Youpeng%22%2C%22lastName%22%3A%22Zhao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chinmay%20Dhanraj%22%2C%22lastName%22%3A%22Nehate%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jun%22%2C%22lastName%22%3A%22Wang%22%7D%5D%2C%22abstractNote%22%3A%22The%20rise%20of%20million-token%2C%20agent-based%20applications%20has%20placed%20unprecedented%20demands%20on%20large%20language%20model%20%28LLM%29%20inference%20services.%20The%20long-running%20nature%20of%20these%20tasks%20increases%20their%20susceptibility%20to%20hardware%20and%20software%20faults%2C%20leading%20to%20costly%20job%20failures%2C%20wasted%20resources%2C%20and%20degraded%20user%20experience.%20The%20stateful%20key-value%20%28KV%29%20cache%2C%20which%20grows%20with%20the%20sequence%20length%2C%20presents%20a%20central%20challenge%20as%20it%20is%20a%20critical%20and%20vulnerable%20component%20in%20distributed%20serving%20systems.%20In%20this%20work%2C%20we%20propose%20GhostServe%2C%20a%20novel%20checkpointing%20solution%20to%20facilitate%20fault-tolerant%20LLM%20serving.%20Specifically%2C%20GhostServe%20protects%20the%20streaming%20KV%20cache%20in%20the%20shadow%20by%20applying%20erasure%20coding%20to%20generate%20and%20store%20the%20parity%20shards%20in%20host%20memory.%20In%20the%20event%20of%20device%20failures%2C%20GhostServe%20enables%20fast%20reconstruction%20of%20the%20lost%20KV%20cache%2C%20allowing%20the%20inference%20process%20to%20resume%20seamlessly%20without%20costly%20full%20recomputation%20or%20state%20replication.%20Evaluations%20demonstrate%20that%20GhostServe%20reduces%20checkpointing%20latency%20by%20up%20to%202.7x%20and%20recovery%20latency%20by%202.1x%20for%20a%20single%20batch%2C%20and%201.2x%20median%20response%20latency%20compared%20to%20existing%20methods%2C%20in%20the%20presence%20of%20system%20failures%2C%20paving%20the%20way%20for%20high-availability%20and%20cost-effective%20LLM%20serving%20at%20scale.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2605.00831%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2605.00831%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T20%3A45%3A29Z%22%7D%7D%2C%7B%22key%22%3A%22PRE2P4N2%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Mo%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-02%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMo%2C%20C.%2C%20Karsin%2C%20B.%2C%20Adams%2C%20P.%20%26amp%3B%20Zhang%2C%20M.%20VecFlow-Chamfer%3A%20A%20GPU-based%20Data%20Management%20System%20for%20High-Performance%20Multi-Vector%20Search%20on%20Superchips.%20%26lt%3Bi%26gt%3BProc.%20ACM%20Manag.%20Data%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B4%26lt%3B%5C%2Fb%26gt%3B%2C%201%26%23x2013%3B26%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22VecFlow-Chamfer%3A%20A%20GPU-based%20Data%20Management%20System%20for%20High-Performance%20Multi-Vector%20Search%20on%20Superchips%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chenghao%22%2C%22lastName%22%3A%22Mo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ben%22%2C%22lastName%22%3A%22Karsin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Philip%22%2C%22lastName%22%3A%22Adams%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Minjia%22%2C%22lastName%22%3A%22Zhang%22%7D%5D%2C%22abstractNote%22%3A%22Many%20emerging%20AI%20applications%20demand%20retrieval%20systems%20that%20go%20beyond%20document-level%20relevance%20and%20capture%20token-level%20semantics.%20Multi-vector%20search%20addresses%20this%20need%20through%20fine-grained%20semantic%20matching%20between%20token-level%20embeddings%20of%20queries%20and%20documents.%20However%2C%20it%20introduces%20significant%20system%20challenges%2C%20including%20compute-intensive%20set-to-set%20scoring%2C%20complex%20candidate%20filtering%2C%20and%20high%20memory%20overhead%20from%20storing%20dense%20token-level%20embeddings.%20Prior%20systems%20mitigate%20these%20challenges%20through%20GPU-based%20similarity%20calculation%20and%20indexing%20structures%20%28e.g.%2C%20IVFPQ-GPU%29%20but%20often%20at%20the%20cost%20of%20reduced%20retrieval%20accuracy%20and%20suffer%20from%20low%20GPU%20utilization.%20We%20present%20%5C%5Cname%2C%20a%20GPU-based%20vector%20data%20management%20system%20that%20enables%20low-latency%20and%20high-recall%20multi-vector%20search%20on%20modern%20Superchip%20architectures.%20%5C%5Cname%20achieves%20this%20through%20a%20combination%20of%20three%20novel%20optimizations%3A%20%281%29%20%5C%5Cmaxivf%2C%20a%20GPU-native%2C%20compression-free%20index%20tailored%20for%20multi-vector%20search%20with%20fine-grained%20anchor%20vectors%20enabling%20scalable%20index%20construction%20and%20low-latency%2C%20high-accuracy%20candidate%20generation%20via%20CAGRA-based%20GPU%20routing%3B%20%282%29%20%5C%5Cchamferkernel%2C%20a%20highly-optimized%20GPU%20kernel%20that%20enables%20single-digit%20millisecond%20Chamfer%20scoring%20over%20tens%20of%20thousands%20of%20candidate%20documents%3B%20and%20%283%29%20%5C%5Czerocomp%2C%20a%20tiered%20vector%20storage%20layer%20that%20supports%20on-demand%2C%20low-latency%20access%20to%20full-precision%20embeddings%20across%20Grace-Hopper%20NVLink-C2C%20interconnects.%20Together%2C%20these%20techniques%20enable%20%5C%5Cname%20to%20perform%20multi-vector%20search%20over%20hundreds%20of%20millions%20of%20document%20token%20embeddings%20with%20unprecedented%20high%20recall%20and%20low%20latency.%20Compared%20to%20state-of-the-art%20systems%20like%20PLAID%20and%20MUVERA%2C%20%5C%5Cname%20achieves%20an%20order-of-magnitude%20lower%20latency%20while%20significantly%20improving%20recall.%22%2C%22date%22%3A%222026-04-02%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1145%5C%2F3786706%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fdl.acm.org%5C%2Fdoi%5C%2F10.1145%5C%2F3786706%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222836-6573%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T19%3A26%3A13Z%22%7D%7D%2C%7B%22key%22%3A%22QVLTYZJI%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Jiang%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-03%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BJiang%2C%20Y.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20DeepDISC-Euclid%3A%20Source%20Classification%20and%20Photometric%20Redshifts%20in%20Euclid%20Deep%20Field%20North%20With%20a%20Pixel-Level%20Deep%20Learning%20Approach.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.03182v2%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.03182v2%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22DeepDISC-Euclid%3A%20Source%20Classification%20and%20Photometric%20Redshifts%20in%20Euclid%20Deep%20Field%20North%20With%20a%20Pixel-Level%20Deep%20Learning%20Approach%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yuanzhe%22%2C%22lastName%22%3A%22Jiang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yue%22%2C%22lastName%22%3A%22Shen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Grant%22%2C%22lastName%22%3A%22Merz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shurui%22%2C%22lastName%22%3A%22Lin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xin%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhiwei%22%2C%22lastName%22%3A%22Pan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mingyang%22%2C%22lastName%22%3A%22Zhuang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%22%2C%22lastName%22%3A%22Roster%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mara%22%2C%22lastName%22%3A%22Salvato%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Malgorzata%22%2C%22lastName%22%3A%22Siudek%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Grant%22%2C%22lastName%22%3A%22Stevens%22%7D%5D%2C%22abstractNote%22%3A%22The%20first%20Euclid%20Quick%20Data%20Release%20%28Q1%29%20provides%20extensive%20imaging%20and%20spectroscopic%20data%20for%20hundreds%20of%20millions%20of%20photometric%20objects%20across%20several%20deep%20fields.%20Accurate%20classifications%20and%20photometric%20redshifts%20%28photo-z%29%20for%20these%20sources%20are%20crucial%20to%20maximizing%20the%20value%20of%20these%20data.%20In%20this%20work%2C%20we%20perform%20source%20classification%20and%20photo-z%20estimation%20for%20the%20Euclid%20Deep%20Field%20North%20%28EDF-N%29%20around%20the%20North%20Ecliptic%20Pole%2C%20using%20a%20deep%20learning%20framework%20%28DeepDISC%29%20that%20learns%20and%20infers%20using%209-band%20images%20simultaneously.%20We%20train%20three%20dedicated%20models%20for%20%281%29%20source%20detection%20and%20classification%2C%20%282%29%20galaxy%20photo-z%2C%20and%20%283%29%20quasar%20photo-z.%20The%20Euclid%20Q1%20input%20source%20catalog%2C%20and%20classifications%20and%20spectroscopic%20redshifts%20%28spec-z%29%20from%20the%20Dark%20Energy%20Spectroscopic%20Instrument%20Data%20Release%201%20are%20adopted%20as%20our%20training%20data.%20DeepDISC%20source%20detection%20achieves%20overall%20completeness%20of%20~93%25%20and%20purity%20of%20~80%25%20if%20using%20the%20Euclid%20source%20catalog%20as%20the%20ground%20truth.%20Using%20a%20JWST%20source%20catalog%20within%20EDF-N%20as%20the%20reference%2C%20we%20estimate%20a%20true%20purity%20of%20~%2090%25%20for%20DeepDISC%20sources.%20About%2099.2%25%2C%2099.0%25%2C%20and%2084.8%25%20of%20stars%2C%20galaxies%2C%20and%20quasars%2C%20respectively%2C%20are%20correctly%20recovered%20with%20their%20spectroscopic%20classifications.%20The%20DeepDISC%20photo-zs%20show%20good%20agreement%20with%20spectroscopic%20redshifts%2C%20for%20both%20galaxies%20and%20quasars.%20Comparisons%20with%20other%20Euclid%20Q1%20products%20demonstrate%20that%20DeepDISC%20provides%20comparable%20or%20improved%20performance%20in%20source%20detection%5C%2Fdeblending%2C%20classification%20and%20photo-z%2C%20especially%20for%20quasars.%20These%20results%20demonstrate%20the%20potential%20of%20pixel-level%20deep%20learning%20approaches%20for%20large-scale%20sky%20surveys%20such%20as%20Euclid%20and%20Roman%2C%20which%20will%20continue%20to%20improve%20with%20better%20training%20labels.%20We%20release%20the%20full%20DeepDISC%20source%20catalog%20%28~13%20million%20objects%29%20for%20EDF-N%20with%20classifications%20and%20photo-zs%2C%20including%20photo-z%20probability%20distributions.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%5C%2F04%5C%2F03%22%2C%22DOI%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.03182v2%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-08T19%3A20%3A56Z%22%7D%7D%2C%7B%22key%22%3A%22S5IBLRPD%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Howes%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHowes%2C%20W.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Graph%20Neural%20Operator%20Towards%20Edge%20Deployability%20and%20Portability%20for%20Sparse-to-Dense%2C%20Real-Time%20Virtual%20Sensing%20on%20Irregular%20Grids.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.01802%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2604.01802%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Graph%20Neural%20Operator%20Towards%20Edge%20Deployability%20and%20Portability%20for%20Sparse-to-Dense%2C%20Real-Time%20Virtual%20Sensing%20on%20Irregular%20Grids%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%22%2C%22lastName%22%3A%22Howes%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jason%22%2C%22lastName%22%3A%22Yoo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kazuma%22%2C%22lastName%22%3A%22Kobayashi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Subhankar%22%2C%22lastName%22%3A%22Sarkar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Farid%22%2C%22lastName%22%3A%22Ahmed%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Souvik%22%2C%22lastName%22%3A%22Chakraborty%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Syed%20Bahauddin%22%2C%22lastName%22%3A%22Alam%22%7D%5D%2C%22abstractNote%22%3A%22Accurate%20sensing%20of%20spatially%20distributed%20physical%20fields%20typically%20requires%20dense%20instrumentation%2C%20which%20is%20often%20infeasible%20in%20real-world%20systems%20due%20to%20cost%2C%20accessibility%2C%20and%20environmental%20constraints.%20Physics-based%20solvers%20address%20this%20through%20direct%20numerical%20integration%20of%20governing%20equations%2C%20but%20their%20computational%20latency%20and%20power%20requirements%20preclude%20real-time%20use%20in%20resource-constrained%20monitoring%20and%20control%20systems.%20Here%20we%20introduce%20VIRSO%20%28Virtual%20Irregular%20Real-Time%20Sparse%20Operator%29%2C%20a%20graph-based%20neural%20operator%20for%20sparse-to-dense%20reconstruction%20on%20irregular%20geometries%2C%20and%20a%20variable-connectivity%20algorithm%2C%20Variable%20KNN%20%28V-KNN%29%2C%20for%20mesh-informed%20graph%20construction.%20Unlike%20prior%20neural%20operators%20that%20treat%20hardware%20deployability%20as%20secondary%2C%20VIRSO%20reframes%20inference%20as%20measurement%3A%20the%20combination%20of%20both%20spectral%20and%20spatial%20analysis%20provides%20accurate%20reconstruction%20without%20the%20high%20latency%20and%20power%20consumption%20of%20previous%20graph-based%20methodologies%20with%20poor%20scalability%2C%20presenting%20VIRSO%20as%20a%20potential%20candidate%20for%20edge-constrained%2C%20real-time%20virtual%20sensing.%20We%20evaluate%20VIRSO%20on%20three%20nuclear%20thermal-hydraulic%20benchmarks%20of%20increasing%20geometric%20and%20multiphysics%20complexity%2C%20across%20reconstruction%20ratios%20from%2047%3A1%20to%20156%3A1.%20VIRSO%20achieves%20mean%20relative%20%24L_2%24%20errors%20below%201%25%2C%20outperforming%20other%20benchmark%20operators%20while%20using%20fewer%20parameters.%20The%20full%2010-layer%20configuration%20reduces%20the%20energy-delay%20product%20%28EDP%29%20from%20%24%7B%5C%5Capprox%7D206%24%20J%24%5C%5Ccdot%24ms%20for%20the%20graph%20operator%20baseline%20to%20%2410.1%24%20J%24%5C%5Ccdot%24ms%20on%20an%20NVIDIA%20H200.%20Implemented%20on%20an%20NVIDIA%20Jetson%20Orin%20Nano%2C%20all%20configurations%20of%20VIRSO%20provide%20sub-10%20W%20power%20consumption%20and%20sub-second%20latency.%20These%20results%20establish%20the%20edge-feasibility%20and%20hardware-portability%20of%20VIRSO%20and%20present%20compute-aware%20operator%20learning%20as%20a%20new%20paradigm%20for%20real-time%20sensing%20in%20inaccessible%20and%20resource-constrained%20environments.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2604.01802%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2604.01802%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T22%3A05%3A23Z%22%7D%7D%2C%7B%22key%22%3A%228EK9ME9U%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Long%20et%20al.%22%2C%22parsedDate%22%3A%222025-09-18%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLong%2C%20D.%2C%20Zhe%2C%20S.%2C%20Williams%2C%20S.%2C%20Oliker%2C%20L.%20%26amp%3B%20Bai%2C%20Z.%20StFT%3A%20Spatio-temporal%20Fourier%20Transformer%20for%20Long-term%20Dynamics%20Prediction.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2503.11899%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2503.11899%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22StFT%3A%20Spatio-temporal%20Fourier%20Transformer%20for%20Long-term%20Dynamics%20Prediction%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Da%22%2C%22lastName%22%3A%22Long%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shandian%22%2C%22lastName%22%3A%22Zhe%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Samuel%22%2C%22lastName%22%3A%22Williams%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Leonid%22%2C%22lastName%22%3A%22Oliker%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zhe%22%2C%22lastName%22%3A%22Bai%22%7D%5D%2C%22abstractNote%22%3A%22Simulating%20the%20long-term%20dynamics%20of%20multi-scale%20and%20multi-physics%20systems%20poses%20a%20significant%20challenge%20in%20understanding%20complex%20phenomena%20across%20science%20and%20engineering.%20The%20complexity%20arises%20from%20the%20intricate%20interactions%20between%20scales%20and%20the%20interplay%20of%20diverse%20physical%20processes%2C%20which%20manifest%20in%20PDEs%20through%20coupled%2C%20nonlinear%20terms%20that%20govern%20the%20evolution%20of%20multiple%20physical%20fields%20across%20scales.%20Neural%20operators%20have%20shown%20potential%20in%20short-term%20prediction%20of%20such%20complex%20spatio-temporal%20dynamics%3B%20however%2C%20achieving%20stable%20high-fidelity%20predictions%20and%20providing%20robust%20uncertainty%20quantification%20over%20extended%20time%20horizons%20remains%20an%20open%20and%20unsolved%20area%20of%20research.%20These%20limitations%20often%20lead%20to%20stability%20degradation%20with%20rapid%20error%20accumulation%2C%20particularly%20in%20long-term%20forecasting%20of%20systems%20characterized%20by%20multi-scale%20behaviors%20involving%20dynamics%20of%20different%20orders.%20To%20address%20these%20challenges%2C%20we%20propose%20an%20autoregressive%20Spatio-temporal%20Fourier%20Transformer%20%28StFT%29%2C%20in%20which%20each%20transformer%20block%20is%20designed%20to%20learn%20the%20system%20dynamics%20at%20a%20distinct%20scale%20through%20a%20dual-path%20architecture%20that%20integrates%20frequency-domain%20and%20spatio-temporal%20representations.%20By%20leveraging%20a%20structured%20hierarchy%20of%20%5C%5Cours%20blocks%2C%20the%20resulting%20model%20explicitly%20captures%20the%20underlying%20dynamics%20across%20both%20macro-%20and%20micro-%20spatial%20scales.%20Furthermore%2C%20a%20generative%20residual%20correction%20mechanism%20is%20introduced%20to%20learn%20a%20probabilistic%20refinement%20temporally%20while%20simultaneously%20quantifying%20prediction%20uncertainties%2C%20enhancing%20both%20the%20accuracy%20and%20reliability%20of%20long-term%20probabilistic%20forecasting.%20Evaluations%20conducted%20on%20three%20benchmark%20datasets%20%28plasma%2C%20fluid%2C%20and%20atmospheric%20dynamics%29%20demonstrate%20the%20advantages%20of%20our%20approach%20over%20state-of-the-art%20ML%20methods.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22arXiv%3A2503.11899%22%2C%22date%22%3A%222025-09-18%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2503.11899%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22http%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2503.11899%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T21%3A56%3A10Z%22%7D%7D%2C%7B%22key%22%3A%22QC65QWFZ%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Zhang%20et%20al.%22%2C%22parsedDate%22%3A%222026-03-30%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BZhang%2C%20Z.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20HandX%3A%20Scaling%20Bimanual%20Motion%20and%20Interaction%20Generation.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2603.28766%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2603.28766%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22HandX%3A%20Scaling%20Bimanual%20Motion%20and%20Interaction%20Generation%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zimu%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yucheng%22%2C%22lastName%22%3A%22Zhang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiyan%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ziyin%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sirui%22%2C%22lastName%22%3A%22Xu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kai%22%2C%22lastName%22%3A%22Zhou%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bing%22%2C%22lastName%22%3A%22Zhou%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chuan%22%2C%22lastName%22%3A%22Guo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jian%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yu-Xiong%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Liang-Yan%22%2C%22lastName%22%3A%22Gui%22%7D%5D%2C%22abstractNote%22%3A%22Synthesizing%20human%20motion%20has%20advanced%20rapidly%2C%20yet%20realistic%20hand%20motion%20and%20bimanual%20interaction%20remain%20underexplored.%20Whole-body%20models%20often%20miss%20the%20fine-grained%20cues%20that%20drive%20dexterous%20behavior%2C%20finger%20articulation%2C%20contact%20timing%2C%20and%20inter-hand%20coordination%2C%20and%20existing%20resources%20lack%20high-fidelity%20bimanual%20sequences%20that%20capture%20nuanced%20finger%20dynamics%20and%20collaboration.%20To%20fill%20this%20gap%2C%20we%20present%20HandX%2C%20a%20unified%20foundation%20spanning%20data%2C%20annotation%2C%20and%20evaluation.%20We%20consolidate%20and%20filter%20existing%20datasets%20for%20quality%2C%20and%20collect%20a%20new%20motion-capture%20dataset%20targeting%20underrepresented%20bimanual%20interactions%20with%20detailed%20finger%20dynamics.%20For%20scalable%20annotation%2C%20we%20introduce%20a%20decoupled%20strategy%20that%20extracts%20representative%20motion%20features%2C%20e.g.%2C%20contact%20events%20and%20finger%20flexion%2C%20and%20then%20leverages%20reasoning%20from%20large%20language%20models%20to%20produce%20fine-grained%2C%20semantically%20rich%20descriptions%20aligned%20with%20these%20features.%20Building%20on%20the%20resulting%20data%20and%20annotations%2C%20we%20benchmark%20diffusion%20and%20autoregressive%20models%20with%20versatile%20conditioning%20modes.%20Experiments%20demonstrate%20high-quality%20dexterous%20motion%20generation%2C%20supported%20by%20our%20newly%20proposed%20hand-focused%20metrics.%20We%20further%20observe%20clear%20scaling%20trends%3A%20larger%20models%20trained%20on%20larger%2C%20higher-quality%20datasets%20produce%20more%20semantically%20coherent%20bimanual%20motion.%20Our%20dataset%20is%20released%20to%20support%20future%20research.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22arXiv%3A2603.28766%22%2C%22date%22%3A%222026-03-30%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2603.28766%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22http%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.28766%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T21%3A50%3A56Z%22%7D%7D%2C%7B%22key%22%3A%22BIT6TAZR%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Guo%20and%20Schwalbe-Koda%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BGuo%2C%20J.%20%26amp%3B%20Schwalbe-Koda%2C%20D.%20Generative%20Inversion%20of%20Spectroscopic%20Data%20for%20Amorphous%20Structure%20Elucidation.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.23210%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.23210%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Generative%20Inversion%20of%20Spectroscopic%20Data%20for%20Amorphous%20Structure%20Elucidation%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jiawei%22%2C%22lastName%22%3A%22Guo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Daniel%22%2C%22lastName%22%3A%22Schwalbe-Koda%22%7D%5D%2C%22abstractNote%22%3A%22Determining%20atomistic%20structures%20from%20characterization%20data%20is%20one%20of%20the%20most%20common%20yet%20intricate%20problems%20in%20materials%20science.%20Particularly%20in%20amorphous%20materials%2C%20proposing%20structures%20that%20balance%20realism%20and%20agreement%20with%20experiments%20requires%20expert%20guidance%2C%20good%20interatomic%20potentials%2C%20or%20both.%20Here%2C%20we%20introduce%20GLASS%2C%20a%20generative%20framework%20that%20inverts%20multi-modal%20spectroscopic%20measurements%20into%20realistic%20atomistic%20structures%20without%20knowledge%20of%20the%20potential%20energy%20surface.%20A%20score-based%20model%20learns%20a%20structural%20prior%20from%20low-fidelity%20data%20and%20samples%20out-of-distribution%20structures%20conditioned%20on%20differentiable%20spectral%20targets.%20Reconstructions%20using%20pair%20distribution%20functions%20%28PDFs%29%2C%20X-ray%20absorption%20spectroscopy%2C%20and%20diffraction%20measurements%20quantify%20the%20complementarity%20between%20spectral%20modalities%20and%20demonstrate%20that%20PDFs%20is%20the%20most%20informative%20probe%20for%20our%20framework.%20We%20use%20GLASS%20to%20rationalize%20three%20contested%20experimental%20problems%3A%20paracrystallinity%20in%20amorphous%20silicon%2C%20a%20liquid-liquid%20phase%20transition%20in%20sulfur%2C%20and%20ball-milled%20amorphous%20ice.%20In%20each%20case%2C%20generated%20structures%20reproduce%20experimental%20measurements%20and%20reveal%20mechanisms%20inaccessible%20to%20diffraction%20analysis%20alone.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.23210%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.23210%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T21%3A45%3A13Z%22%7D%7D%2C%7B%22key%22%3A%22N8UZBHXN%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Cruz-Camacho%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BCruz-Camacho%2C%20N.%2C%20Conde-Ocazionez%2C%20C.%2C%20Dexheimer%2C%20V.%2C%20Noronha-Hostler%2C%20J.%20%26amp%3B%20Yunes%2C%20N.%20Sensitivity%20of%20neutron%20star%20observables%20to%20microscopic%20nuclear%20parameters%20of%20realistic%20equations%20of%20state.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.16019%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.16019%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Sensitivity%20of%20neutron%20star%20observables%20to%20microscopic%20nuclear%20parameters%20of%20realistic%20equations%20of%20state%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nikolas%22%2C%22lastName%22%3A%22Cruz-Camacho%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Carlos%22%2C%22lastName%22%3A%22Conde-Ocazionez%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Veronica%22%2C%22lastName%22%3A%22Dexheimer%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jacquelyn%22%2C%22lastName%22%3A%22Noronha-Hostler%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nicol%5Cu00e1s%22%2C%22lastName%22%3A%22Yunes%22%7D%5D%2C%22abstractNote%22%3A%22The%20equation%20of%20state%20of%20matter%20at%20supranuclear%20densities%20governs%20the%20astrophysical%20observables%20of%20neutron%20stars.%20A%20realistic%2C%20though%20complex%2C%20description%20is%20provided%20by%20the%20Chiral-Mean-Field%20model%2C%20which%20depends%20on%20many%20microscopic%20nuclear-physics%20parameters.%20We%20present%20a%20Fisher-information-inspired%20analysis%20of%20the%20sensitivity%20of%20neutron-star%20observables%20to%20the%20parameters%20of%20the%20Chiral-Mean-Field%20model%20at%20%24%5Cu03b2%24-equilibrium%20using%20SLy%20as%20a%20crust.%20We%20then%20compute%20neutron-star%20sequences%20and%20extract%20masses%2C%20radii%2C%20compactnesses%2C%20and%20tidal%20deformabilities.%20From%20the%20logarithmic%20derivatives%20of%20these%20observables%20with%20respect%20to%20each%20nuclear%20parameter%2C%20we%20construct%20a%20dimensionless%2C%20Fisher-inspired%20sensitivity%20matrix%20and%20perform%20a%20principal-component%20analysis%20to%20identify%20the%20effective%20combinations%20of%20nuclear%20parameters%20that%20most%20strongly%20affect%20neutron-star%20observables.%20Although%20the%20ranking%20depends%20mildly%20on%20the%20observable%2C%20the%20three%20most%20important%20nuclear%20parameters%20are%20the%20vacuum%20value%20of%20the%20dilaton%20field%20%24%5Cu03c7_0%24%20%28which%20sets%20the%20overall%20scale%20of%20the%20scalar%20potential%20and%20trace-anomaly%20contribution%29%2C%20the%20scalar%20singlet%20strength%20%24g_%7B1%7D%5EX%24%20%28which%20controls%20the%20overall%20scalar%20attraction%20through%20the%20baryon%20effective%20masses%29%2C%20and%20the%20%24k_0%24%20quadratic%20scalar%20term%20%28which%20governs%20the%20curvature%20of%20the%20scalar%20potential%29.%20This%20framework%20provides%20a%20reproducible%2C%20data-driven%20approach%20to%20quantify%20parameter%20sensitivities%20in%20dense-matter%20models%20and%20to%20guide%20future%20Bayesian%20inference%20of%20nuclear%20information%20from%20multi-messenger%20astrophysical%20observations.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.16019%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.16019%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T21%3A41%3A32Z%22%7D%7D%2C%7B%22key%22%3A%22R3V48UIN%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lihan%20et%20al.%22%2C%22parsedDate%22%3A%222026-04-30%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLihan%2C%20M.%2C%20Pant%2C%20S.%2C%20Ojoawo%2C%20A.%20M.%2C%20Rienstra%2C%20C.%20M.%20%26amp%3B%20Tajkhorshid%2C%20E.%20Probing%20Solution%20Dynamics%20of%20Tissue%20Factor%20Using%20Molecular%20Dynamics%20Simulations%20Guided%20by%20NMR%20Chemical%20Shifts.%20%26lt%3Bi%26gt%3BJ.%20Phys.%20Chem.%20B%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B130%26lt%3B%5C%2Fb%26gt%3B%2C%204528%26%23x2013%3B4537%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Probing%20Solution%20Dynamics%20of%20Tissue%20Factor%20Using%20Molecular%20Dynamics%20Simulations%20Guided%20by%20NMR%20Chemical%20Shifts%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Muyun%22%2C%22lastName%22%3A%22Lihan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shashank%22%2C%22lastName%22%3A%22Pant%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Adedolapo%20M.%22%2C%22lastName%22%3A%22Ojoawo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chad%20M.%22%2C%22lastName%22%3A%22Rienstra%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Emad%22%2C%22lastName%22%3A%22Tajkhorshid%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22date%22%3A%222026-04-30%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1021%5C%2Facs.jpcb.6c00315%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fpubs.acs.org%5C%2Fdoi%5C%2F10.1021%5C%2Facs.jpcb.6c00315%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%221520-6106%2C%201520-5207%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T21%3A20%3A18Z%22%7D%7D%2C%7B%22key%22%3A%22DQVJFZ25%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Li%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLi%2C%20C.%2C%20Merkel%2C%20M.%20%26amp%3B%20Sussman%2C%20D.%20M.%20Models%20of%203D%20confluent%20tissue%20as%20under-constrained%20glasses.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.24438%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.24438%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Models%20of%203D%20confluent%20tissue%20as%20under-constrained%20glasses%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Chengling%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Matthias%22%2C%22lastName%22%3A%22Merkel%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Daniel%20M.%22%2C%22lastName%22%3A%22Sussman%22%7D%5D%2C%22abstractNote%22%3A%22The%20dynamics%20of%20glassy%20materials%20slows%20down%20upon%20cooling%2C%20typically%20showing%20either%20Arrhenius%20or%20super-Arrhenius%20behavior.%20However%2C%20it%20was%20recently%20shown%20that%202D%20cell-based%20models%20for%20biological%20tissues%20can%20be%20continuously%20tuned%20between%20Arrhenius%20and%20sub-Arrhenius%20dynamics.%20In%20previous%20work%2C%20using%20the%202D%20Voronoi%20model%2C%20we%20proposed%20that%20such%20atypical%20dynamical%20behavior%20could%20be%20a%20generic%20feature%20of%20the%20broad%20class%20of%20mechanically%20under-constrained%20materials.%20Our%20earlier%20study%20had%20left%20two%20important%20points%20open%3A%20%281%29%20many%202D%20systems%20are%20affected%20by%20long-wavelength%20fluctuations%20and%20the%202D%20melting%20scenario%2C%20and%20%282%29%20the%202D%20Voronoi%20model%20sits%20exactly%20at%20the%20isostatic%20point%2C%20making%20it%20a%20marginal%20case%20rather%20than%20a%20strictly%20under-constrained%20one.%20Both%20points%20complicate%20the%20interpretation%20of%20our%202D%20Voronoi%20model%20results%20and%20their%20generalization%20to%20other%20systems%3B%20to%20remedy%20this%2C%20here%20we%20use%20large-scale%20simulations%20to%20study%20the%20glassy%20behavior%20of%20the%203D%20extension%20of%20the%20Voronoi%20model.%20We%20first%20show%20that%20the%20structural%20relaxation%20time%20%24%5Cu03c4_%5Cu03b1%24%20of%20the%203D%20Voronoi%20model%20can%20be%20tuned%20between%20sub-Arrhenius%20and%20Arrhenius%20behavior%2C%20like%20the%202D%20Voronoi%20model.%20We%20then%20establish%20that%20the%20four-point%20susceptibility%2C%20the%20structure%20factor%2C%20and%20the%20model%26%23039%3Bs%20mechanical%20properties%20all%20display%20trends%20consistent%20with%20the%202D%20Voronoi%20model.%20These%20results%20provide%20strong%20evidence%20that%20sub-Arrhenius%20glassy%20dynamics%20are%20a%20generic%20feature%20of%20under-constrained%20materials%20across%20dimensions.%20Our%20work%20thus%20broadens%20the%20class%20of%20disordered%20materials%20known%20to%20have%20highly%20unusual%20glassy%20phenomenology.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.24438%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.24438%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T21%3A11%3A00Z%22%7D%7D%2C%7B%22key%22%3A%22MX6ZBVUY%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Cassar%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BCassar%2C%20N.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20%24B%20%5C%5Cto%20%26%23x3C0%3B%24%2C%20%24B_%7B%28s%29%7D%20%5C%5Cto%20D_%7B%28s%29%7D%24%20from%202%2B1%2B1%20Flavor%20Lattice%20QCD.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.23702%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.23702%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22%24B%20%5C%5Cto%20%5Cu03c0%24%2C%20%24B_%7B%28s%29%7D%20%5C%5Cto%20D_%7B%28s%29%7D%24%20from%202%2B1%2B1%20Flavor%20Lattice%20QCD%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nicholas%22%2C%22lastName%22%3A%22Cassar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Akhil%22%2C%22lastName%22%3A%22Chauhan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Carleton%22%2C%22lastName%22%3A%22DeTar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aida%22%2C%22lastName%22%3A%22El-Khadra%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Elvira%22%2C%22lastName%22%3A%22G%5Cu00e1miz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Steven%22%2C%22lastName%22%3A%22Gottlieb%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22William%20I.%22%2C%22lastName%22%3A%22Jay%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andreas%20S.%22%2C%22lastName%22%3A%22Kronfeld%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jack%22%2C%22lastName%22%3A%22Laiho%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andrew%22%2C%22lastName%22%3A%22Lytle%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alejandro%22%2C%22lastName%22%3A%22Vaquero%22%7D%5D%2C%22abstractNote%22%3A%22We%20present%20a%20lattice-QCD%20calculation%20of%20the%20hadronic%20form%20factors%20for%20%24B%24-meson%20semileptonic%20decays%20computed%20using%20the%20highly%20improved%20staggered%20quark%20action%20for%20both%20valence%20and%20sea%20quarks%20on%20the%20MILC%20collaborations%202%2B1%2B1-flavor%20ensembles%20with%20lattice%20spacing%20ranging%20from%200.09%20fm%20to%200.03%20fm%2C%20many%20with%20physical%20pion%20masses%20On%20our%20finest%20ensembles%2C%20we%20compute%20the%20form%20factors%20directly%20at%20the%20physical%20%24b%24-quark%20mass.%20We%20discuss%20the%20computational%20setup%20and%20analysis%20strategies%20for%20two-%20and%20three-point%20correlation%20functions.%20For%20%24B_%7B%28s%29%7D%20%5C%5Cto%20D_%7B%28s%29%7D%24%20we%20present%20preliminary%20results%20of%20chiral-continuum%20fits%20for%20the%20scalar%20and%20vector%20form%20factors.%20The%20goal%20of%20this%20project%20is%20a%20percent-level%20determination%20of%20the%20scalar%20and%20vector%20form%20factors%20to%20enable%20high-precision%20determinations%20of%20%24%7CV_%7Bub%7D%7C%24%20and%20%24%7CV_%7Bcb%7D%7C%24.%20This%20work%20fits%20into%20a%20broader%20program%20of%20lattice-QCD%20studies%20of%20weak%20%24B%24-meson%20decays%20by%20the%20Fermilab%20Lattice%20and%20MILC%20Collaborations.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.23702%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.23702%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T20%3A37%3A17Z%22%7D%7D%2C%7B%22key%22%3A%22AJ7V97BG%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lin%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLin%2C%20S.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20D%24_4%24CNN%24%5C%5Ctimes%24AnaCal%3A%20Physics-Informed%20Machine%20Learning%20for%20Accurate%20and%20Precise%20Weak%20Lensing%20Shear%20Estimation.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.19046%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.19046%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22D%24_4%24CNN%24%5C%5Ctimes%24AnaCal%3A%20Physics-Informed%20Machine%20Learning%20for%20Accurate%20and%20Precise%20Weak%20Lensing%20Shear%20Estimation%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shurui%22%2C%22lastName%22%3A%22Lin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiangchong%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ji%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shengcao%22%2C%22lastName%22%3A%22Cao%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xin%22%2C%22lastName%22%3A%22Liu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yu-Xiong%22%2C%22lastName%22%3A%22Wang%22%7D%5D%2C%22abstractNote%22%3A%22Traditional%20weak%20gravitational%20lensing%20shear%20estimators%20are%20carefully%20calibrated%20but%20struggle%20to%20fully%20capture%20realistic%20galaxy%20morphologies%2C%20point-spread-function%20%28PSF%29%20effects%2C%20blending%2C%20and%20noise%20in%20deep%20surveys%2C%20while%20blindly%20trained%20machine%20learning%20%28ML%29%20models%20can%20introduce%20significant%20calibration%20biases.%20Here%20we%20construct%20a%20fully%20D%24_4%24-equivariant%20deep%20neural%20network%20for%20galaxy%20shape%20measurement%20whose%20architecture%20enforces%20symmetry%20under%2090%24%5E%7B%5C%5Ccirc%7D%24%20rotations%20and%20mirror%20transformations%2C%20and%20adopt%20the%20Analytical%20Calibration%20framework%20%28AnaCal%29%20to%20calibrate%20the%20model%20using%20its%20backpropagated%20gradients.%20For%20isolated%20galaxies%20in%20LSST-like%20single-band%20simulations%2C%20we%20demonstrate%20that%20our%20approach%20achieves%20%24%5C%5Csim%2410%25%20lower%20shape%20noise%20than%20the%20traditional%20moment-based%20Fourier%20Power%20Function%20Shapelets%20estimator%20in%20the%20high-noise%20regime%2C%20equivalent%20to%20a%20%24%5C%5Csim%2420%25%20gain%20in%20effective%20galaxy%20number%20density%2C%20while%20simultaneously%20achieving%20multiplicative%20biases%20consistent%20with%20zero%20across%20a%20wide%20range%20of%20noise%20levels%2C%20PSF%20sizes%20and%20ellipticities%2C%20and%20magnitude%20selection%20cuts%2C%20with%20all%20measurements%20satisfying%20%24%7Cm%7C%20%7B%26lt%3B%7D%2010%5E%7B-3%7D%24%20%28i.e.%2C%20within%20the%200.2%25%20LSST%20requirement%29%20and%20most%20at%20the%20%24%7B%5C%5Csim%7D10%5E%7B-4%7D%24%20level.%20We%20demonstrate%20this%20framework%20on%20isolated%20single-band%20galaxy%20images%20with%20Gaussian%20noise%20and%20known%20PSF%2C%20establishing%20a%20rigorous%2C%20physics-informed%20foundation%20for%20future%20extensions%20of%20ML-based%20shear%20estimation%20to%20blended%20sources%20and%20multi-band%20observations%20in%20Stage-IV%20surveys.%20All%20codes%20and%20data%20products%20will%20be%20made%20publicly%20available%20upon%20acceptance.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.19046%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.19046%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T20%3A23%3A14Z%22%7D%7D%2C%7B%22key%22%3A%22YBM7SL8H%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lazarian%20et%20al.%22%2C%22parsedDate%22%3A%222025%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLazarian%2C%20A.%2C%20Pogosyan%2C%20D.%20%26amp%3B%20Hu%2C%20Y.%20Model%20of%20super-Alfv%26%23xE9%3Bnic%20MHD%20turbulence%20and%20structure%20functions%20of%20polarization.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2511.08800%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2511.08800%26lt%3B%5C%2Fa%26gt%3B%20%282025%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Model%20of%20super-Alfv%5Cu00e9nic%20MHD%20turbulence%20and%20structure%20functions%20of%20polarization%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22A.%22%2C%22lastName%22%3A%22Lazarian%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22D.%22%2C%22lastName%22%3A%22Pogosyan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Y.%22%2C%22lastName%22%3A%22Hu%22%7D%5D%2C%22abstractNote%22%3A%22MHD%20turbulence%20driven%20at%20velocities%20higher%20than%20the%20Alfv%5Cu00e9n%20velocity%2C%20i.e.%2C%20super-Alfv%5Cu00e9nic%20turbulence%2C%20is%20widely%20spread%20in%20astrophysical%20environments%2C%20including%20galaxy%20clusters%20and%20molecular%20clouds.%20For%20statistical%20studies%20of%20such%20turbulence%2C%20we%20explore%20the%20utility%20of%20the%20polarization%20angle%20structure%20functions%20%24D%5E%5Cu03d5%28R%29%3D%20%5C%5Cleft%5C%5Clangle%5C%5Csin%5E2%28%5Cu03d5_1-%5Cu03d5_2%29%20%5C%5Cright%5C%5Crangle%24%2C%20where%20%24%5Cu03d5%24%20denotes%20the%20polarization%20angle%20measured%20at%20points%20separated%20by%20a%20projected%20distance%20%24%5C%5Cmathbf%7BR%7D%24%20on%20the%20plane%20of%20the%20sky.%20Lazarian%2C%20Yuen%20and%20Pogosyan%2C%202022%2C%20showed%20that%20in%20the%20case%20of%20super-Alfv%5Cu00e9nic%20turbulence%2C%20the%20spectral%20slope%20of%20%24D%5E%5Cu03d5%28%5C%5Cmathbf%7BR%7D%29%24%20differs%20from%20that%20of%20the%20underlying%20magnetic%20fluctuations%2C%20limiting%20its%20applicability%20for%20field%20strength%20estimation%20with%20known%20techniques.%20In%20this%20work%2C%20we%20provide%20an%20analytical%20framework%20that%20explains%20the%20modification%20of%20the%20%24D%5E%5Cu03d5%28R%29%24%20spectral%20slope%20in%20super-Alfv%5Cu00e9nic%20turbulence%20and%20validate%20our%20predictions%20with%20numerical%20simulations.%20We%20demonstrate%20that%20for%20super-Alfv%5Cu00e9nic%20turbulence%2C%20the%20structure%20function%20%24D%5E%5Cu03d5%28R%29%24%20gets%20shallower%20with%20the%20increase%20of%20%24M_A%24.%20Our%20study%20makes%20%24D%5E%5Cu03d5%28R%29%24%20a%20valuable%20diagnostic%20of%20super-Alfv%5Cu00e9nic%20turbulence%20and%20opens%20a%20way%20to%20obtain%20%24M_A%24%20from%20observations.%20We%20also%20explore%20numerically%20the%20structure%20function%20of%20the%20polarization%20degree%20and%20the%20spectrum%20of%20the%20polarization%20directions%2C%20the%20latter%20being%20the%20Fourier%20transform%20of%20%24D%5E%5Cu03d5%24.%20We%20discuss%20the%20implications%20of%20our%20findings%20for%20turbulence%20and%20magnetic%20field%20studies%20in%20the%20intracluster%20and%20interstellar%20media.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222025%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2511.08800%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2511.08800%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-05-01T20%3A18%3A39Z%22%7D%7D%2C%7B%22key%22%3A%22I6QUBWXA%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Li%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLi%2C%20X.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20MCLR%3A%20Improving%20Conditional%20Modeling%20via%20Inter-Class%20Likelihood-Ratio%20Maximization%20and%20Unifying%20Classifier-Free%20Guidance%20with%20Alignment%20Objectives.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.22364%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.22364%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22MCLR%3A%20Improving%20Conditional%20Modeling%20via%20Inter-Class%20Likelihood-Ratio%20Maximization%20and%20Unifying%20Classifier-Free%20Guidance%20with%20Alignment%20Objectives%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiang%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Yixuan%22%2C%22lastName%22%3A%22Jia%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Xiao%22%2C%22lastName%22%3A%22Li%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jeffrey%20A.%22%2C%22lastName%22%3A%22Fessler%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rongrong%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Qing%22%2C%22lastName%22%3A%22Qu%22%7D%5D%2C%22abstractNote%22%3A%22Diffusion%20models%20have%20achieved%20state-of-the-art%20performance%20in%20generative%20modeling%2C%20but%20their%20success%20often%20relies%20heavily%20on%20classifier-free%20guidance%20%28CFG%29%2C%20an%20inference-time%20heuristic%20that%20modifies%20the%20sampling%20trajectory.%20From%20a%20theoretical%20perspective%2C%20diffusion%20models%20trained%20with%20standard%20denoising%20score%20matching%20%28DSM%29%20are%20expected%20to%20recover%20the%20target%20data%20distribution%2C%20raising%20the%20question%20of%20why%20inference-time%20guidance%20is%20necessary%20in%20practice.%20In%20this%20work%2C%20we%20ask%20whether%20the%20DSM%20training%20objective%20can%20be%20modified%20in%20a%20principled%20manner%20such%20that%20standard%20reverse-time%20sampling%2C%20without%20inference-time%20guidance%2C%20yields%20effects%20comparable%20to%20CFG.%20We%20identify%20insufficient%20inter-class%20separation%20as%20a%20key%20limitation%20of%20standard%20diffusion%20models.%20To%20address%20this%2C%20we%20propose%20MCLR%2C%20a%20principled%20alignment%20objective%20that%20explicitly%20maximizes%20inter-class%20likelihood-ratios%20during%20training.%20Models%20fine-tuned%20with%20MCLR%20exhibit%20CFG-like%20improvements%20under%20standard%20sampling%2C%20achieving%20comparable%20qualitative%20and%20quantitative%20gains%20without%20requiring%20inference-time%20guidance.%20Beyond%20empirical%20benefits%2C%20we%20provide%20a%20theoretical%20result%20showing%20that%20the%20CFG-guided%20score%20is%20exactly%20the%20optimal%20solution%20to%20a%20weighted%20MCLR%20objective.%20This%20establishes%20a%20formal%20equivalence%20between%20classifier-free%20guidance%20and%20alignment-based%20objectives%2C%20offering%20a%20mechanistic%20interpretation%20of%20CFG.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.22364%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.22364%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-21T14%3A13%3A43Z%22%7D%7D%2C%7B%22key%22%3A%22KCHCHIGC%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wang%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWang%2C%20S.%2C%20Kang%2C%20W.%20%26amp%3B%20Li%2C%20C.%20A%20Tug-of-War%20Between%20Baroclinic%20Eddies%20and%20Convection%3A%20Implications%20for%20Icy%20Moon%20Oceans.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.17185%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.17185%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22A%20Tug-of-War%20Between%20Baroclinic%20Eddies%20and%20Convection%3A%20Implications%20for%20Icy%20Moon%20Oceans%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shuang%22%2C%22lastName%22%3A%22Wang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Wanying%22%2C%22lastName%22%3A%22Kang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Cheng%22%2C%22lastName%22%3A%22Li%22%7D%5D%2C%22abstractNote%22%3A%22In%20many%20geophysical%20and%20planetary%20environments%2C%20such%20as%20Earth%26%23039%3Bs%20ocean%20and%20atmosphere%20as%20well%20as%20subsurface%20oceans%20of%20icy%20satellites%2C%20convection%20driven%20by%20bottom%20geothermal%20heating%20usually%20coexists%20with%20baroclinic%20eddies%20driven%20by%20lateral%20buoyancy%5C%2Ftemperature%20gradients.%20These%20processes%20compete%20against%20each%20other%2C%20with%20convection%20destabilizing%20the%20stratification%20and%20baroclinic%20eddies%20re-stabilizing%20it%2C%20thereby%20controlling%20whether%20the%20bottom%20heat%20flux%20is%20significantly%20redistributed%20as%20it%20is%20transmitted%20to%20the%20upper%20surface.%20Using%20scaling%20analysis%20and%20numerical%20simulations%2C%20we%20show%20that%20a%20stratified%20layer%20persists%20near%20the%20upper%20surface%20up%20to%20%24%7B%5C%5Crm%20Ra%7D_%7Bv%7D%5C%5Csim%20%7B%5C%5Crm%20Ra%7D_h%5E%7B5%5C%2F2%7D%24%2C%20where%20%24%7B%5C%5Crm%20Ra%7D_h%5C%5Cequiv%20%5Cu0394b_0%5C%2F%28L_zf%5E2%29%24%20measures%20the%20imposed%20upper-surface%20buoyancy%20contrast%20%24%5Cu0394b_0%24%20and%20%24%7B%5C%5Crm%20Ra%7D_v%5C%5Cequiv%20B_0%5C%2F%28L_z%5E2f%5E3%29%24%20measures%20the%20strength%20of%20the%20bottom%20buoyancy%20flux%20%24B_0%24%2C%20%24L_z%24%20is%20the%20domain%20depth%20and%20%24f%24%20is%20the%20Coriolis%20parameter.%20For%20%24%7B%5C%5Crm%20Ra%7D_v%26lt%3B%7B%5C%5Crm%20Ra%7D_h%5E%7B5%5C%2F2%7D%24%2C%20baroclinic%20eddies%20dominate%20over%20convection%2C%20maintain%20the%20upper%20stratified%20layer%2C%20and%20completely%20deflect%20the%20bottom%20buoyancy%5C%2Fheat%20input%20into%20meridional%20transport.%20In%20contrast%2C%20when%20%24%7B%5C%5Crm%20Ra%7D_v%26gt%3B%7B%5C%5Crm%20Ra%7D_h%5E%7B5%5C%2F2%7D%24%2C%20convective%20plumes%20penetrate%20the%20stratification%20and%20transport%20buoyancy%5C%2Fheat%20vertically%20with%20negligible%20deflection.%20Building%20on%20these%20results%2C%20we%20further%20propose%20a%20scaling%20law%20for%20the%20meridional%20buoyancy%5C%2Fheat%20transport%20in%20this%20system.%20Applications%20to%20icy%20satellites%20are%20discussed.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.17185%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.17185%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T22%3A20%3A20Z%22%7D%7D%2C%7B%22key%22%3A%22ES4RL8J4%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Meher%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BMeher%2C%20S.%20Build%2C%20Borrow%2C%20or%20Just%20Fine-Tune%3F%20A%20Political%20Scientist%26%23x2019%3Bs%20Guide%20to%20Choosing%20NLP%20Models.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.09595%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.09595%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Build%2C%20Borrow%2C%20or%20Just%20Fine-Tune%3F%20A%20Political%20Scientist%27s%20Guide%20to%20Choosing%20NLP%20Models%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shreyas%22%2C%22lastName%22%3A%22Meher%22%7D%5D%2C%22abstractNote%22%3A%22Political%20scientists%20increasingly%20face%20a%20consequential%20choice%20when%20adopting%20natural%20language%20processing%20tools%3A%20build%20a%20domain-specific%20model%20from%20scratch%2C%20borrow%20and%20adapt%20an%20existing%20one%2C%20or%20simply%20fine-tune%20a%20general-purpose%20model%20on%20task%20data%3F%20Each%20approach%20occupies%20a%20different%20point%20on%20the%20spectrum%20of%20performance%2C%20cost%2C%20and%20required%20expertise%2C%20yet%20the%20discipline%20has%20offered%20little%20empirical%20guidance%20on%20how%20to%20navigate%20this%20trade-off.%20This%20paper%20provides%20such%20guidance.%20Using%20conflict%20event%20classification%20as%20a%20test%20case%2C%20I%20fine-tune%20ModernBERT%20on%20the%20Global%20Terrorism%20Database%20%28GTD%29%20to%20create%20Confli-mBERT%20and%20systematically%20compare%20it%20against%20ConfliBERT%2C%20a%20domain-specific%20pretrained%20model%20that%20represents%20the%20current%20gold%20standard.%20Confli-mBERT%20achieves%2075.46%25%20accuracy%20compared%20to%20ConfliBERT%26%23039%3Bs%2079.34%25.%20Critically%2C%20the%20four-percentage-point%20gap%20is%20not%20uniform%3A%20on%20high-frequency%20attack%20types%20such%20as%20Bombing%5C%2FExplosion%20%28F1%20%3D%200.95%20vs.%200.96%29%20and%20Kidnapping%20%28F1%20%3D%200.92%20vs.%200.91%29%2C%20the%20models%20are%20nearly%20indistinguishable.%20Performance%20differences%20concentrate%20in%20rare%20event%20categories%20comprising%20fewer%20than%202%25%20of%20all%20incidents.%20I%20use%20these%20findings%20to%20develop%20a%20practical%20decision%20framework%20for%20political%20scientists%20considering%20any%20NLP-assisted%20research%20task%3A%20when%20does%20the%20research%20question%20demand%20a%20specialized%20model%2C%20and%20when%20does%20an%20accessible%20fine-tuned%20alternative%20suffice%3F%20The%20answer%2C%20I%20argue%2C%20depends%20not%20on%20which%20model%20is%20%26quot%3Bbetter%26quot%3B%20in%20the%20abstract%2C%20but%20on%20the%20specific%20intersection%20of%20class%20prevalence%2C%20error%20tolerance%2C%20and%20available%20resources.%20The%20model%2C%20training%20code%2C%20and%20data%20are%20publicly%20available%20on%20Hugging%20Face.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.09595%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.09595%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T22%3A12%3A54Z%22%7D%7D%2C%7B%22key%22%3A%22M93H6PA7%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Ung%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BUng%2C%20S.%20F.%2C%20Mahajan%2C%20A.%20%26amp%3B%20Reichman%2C%20D.%20R.%20Study%20of%20the%20triangular-lattice%20Hubbard%20model%20with%20constrained-path%20quantum%20Monte%20Carlo.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.14808%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.14808%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Study%20of%20the%20triangular-lattice%20Hubbard%20model%20with%20constrained-path%20quantum%20Monte%20Carlo%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Shu%20Fay%22%2C%22lastName%22%3A%22Ung%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ankit%22%2C%22lastName%22%3A%22Mahajan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22David%20R.%22%2C%22lastName%22%3A%22Reichman%22%7D%5D%2C%22abstractNote%22%3A%22We%20benchmark%20constrained-path%20Monte%20Carlo%20%28CPMC%29%20on%20the%20triangular-lattice%20Hubbard%20model%20for%20several%20fillings%20and%20%24U%24%20values%20and%20show%20that%20symmetry-adapted%20trial%20wave%20functions%20are%20essential%20for%20quantitative%20accuracy.%20Away%20from%20half-filling%2C%20simple%20free-electron-based%20trials%20that%20preserve%20the%20ground%20state%20symmetry%20yield%20energy%20deviations%20%24%5C%5Clesssim%201%5C%5C%25%24%20from%20exact%20diagonalization%20and%20density%20matrix%20renormalization%20group%20results.%20At%20half-filling%2C%20strong%20frustration%20in%20the%20intermediate%20to%20large%20%24U%24%20regimes%20necessitates%20symmetry-projected%20trials%20to%20reach%20comparable%20accuracy%2C%20where%20both%20free-electron%20and%20symmetry-broken%20Hartree-Fock%20trials%20incur%20substantial%20constraint%20bias.%20Since%20the%20computational%20cost%20of%20CPMC%20with%20symmetry%20projection%20scales%20polynomially%20with%20system%20size%2C%20our%20results%20motivate%20its%20use%20as%20a%20practical%20route%20for%20studying%20competing%20ground%20states%20in%20strongly%20correlated%2C%20frustrated%20systems.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.14808%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.14808%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T22%3A11%3A13Z%22%7D%7D%2C%7B%22key%22%3A%22DHNKNJK7%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Wong%20and%20Yang%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BWong%2C%20N.%20%26amp%3B%20Yang%2C%20J.%20H.%20Bias%20in%20Universal%20Machine-Learned%20Interatomic%20Potentials%20and%20its%20Effects%20on%20Fine-Tuning.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.10159%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.10159%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Bias%20in%20Universal%20Machine-Learned%20Interatomic%20Potentials%20and%20its%20Effects%20on%20Fine-Tuning%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Nicolas%22%2C%22lastName%22%3A%22Wong%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Julia%20H.%22%2C%22lastName%22%3A%22Yang%22%7D%5D%2C%22abstractNote%22%3A%22Universal%20machine%20learned%20interatomic%20potentials%20%28uMLIPs%29%20embody%20a%20growing%20area%20of%20interest%20due%20to%20their%20transferability%20across%20the%20periodic%20table%2C%20displaying%20an%20error%20of%20about%200.6%20kcal%5C%2Fmol%20against%20the%20Matbench%20Discovery%20test%20set.%20However%2C%20we%20show%20that%20achieving%20more%20accurate%20predictions%20on%20out-of-domain%20tasks%20requires%20fine-tuning.%20Additionally%2C%20we%20investigate%20the%20existence%20and%20influence%20of%20model%20biases%20in%20molecular%20dynamics%20%28MD%29%20by%20examining%20two%20approaches%20for%20data%20generation%3A%20from%20multiple%20MD%20trajectories%20in%20parallel%2C%20which%20we%20call%20naive%20fine-tuning%2C%20and%20from%20a%20single%20MD%20trajectory%20with%20fine-tuning%20after%20set%20intervals%2C%20which%20we%20call%20periodic%20fine-tuning.%20Our%20results%20find%20that%20naive%20fine-tuning%20generates%20constrained%20datasets%20that%20fail%20to%20represent%20MD%20simulations%2C%20and%20thus%20downstream%20fine-tuned%20models%20fail%20during%20extrapolation.%20In%20contrast%2C%20periodic%20fine-tuning%20yields%20models%20which%20are%20more%20generalizable%20and%20accurate%2C%20producing%20low-error%20dynamics.%20These%20findings%20indicate%20the%20role%20of%20uMLIP%20bias%20in%20fine-tuning%2C%20and%20highlights%20the%20need%20for%20multiple%20fine-tuning%20steps.%20Lastly%2C%20we%20relate%20unphysical%20behavior%20to%20principal%20component%20space%2C%20and%20quantify%20extrapolations%20through%20Q-residual%20analysis%2C%20which%20are%20useful%20as%20a%20proxy%20for%20epistemic%20uncertainty%20for%20larger%20simulations.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.10159%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.10159%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T22%3A07%3A06Z%22%7D%7D%2C%7B%22key%22%3A%22HLUV3S5A%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Linser%20et%20al.%22%2C%22parsedDate%22%3A%222026-03-09%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLinser%2C%20R.%2C%20Acharyya%2C%20S.%2C%20Weisner%2C%20J.%20%26amp%3B%20Bernardi%2C%20R.%20Mapping%20of%20dynamic%20allostery%20within%20p38%20alpha%20kinase%20via%20network%20analyses%20and%20NMR%20spectroscopy.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.21203%5C%2Frs.3.rs-8958159%5C%2Fv1%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.21203%5C%2Frs.3.rs-8958159%5C%2Fv1%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Mapping%20of%20dynamic%20allostery%20within%20p38%20alpha%20kinase%20via%20network%20analyses%20and%20NMR%20spectroscopy%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rasmus%22%2C%22lastName%22%3A%22Linser%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Suchandra%22%2C%22lastName%22%3A%22Acharyya%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22J%5Cu00f6rn%22%2C%22lastName%22%3A%22Weisner%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rafael%22%2C%22lastName%22%3A%22Bernardi%22%7D%5D%2C%22abstractNote%22%3A%22Abstract%20%5Cn%20%20%20%20%20%20%20%20%20%20Kinases%20are%20major%20drug%20targets%20especially%20in%20cancer%20therapy.%20However%2C%20the%20high%20degree%20of%20conservation%20of%20their%20active%20sites%20hinders%20the%20development%20of%20selective%20inhibitors%2C%20motivating%20a%20deeper%20understanding%20of%20kinase%20conformational%20ensembles%20and%20allosteric%20communication%20pathways.%20Here%2C%20we%20use%20dynamical%20network%20analysis%20to%20identify%20key%20residues%20involved%20in%20a%20dynamic%20allostery%20between%20the%20N-%20and%20C-lobes%20that%20connects%20the%20major%20functional%20units%20of%20the%20MAP%20kinase%20p38%5Cu03b1.%20By%20combining%20NMR%20spectroscopy%2C%20activity%20assays%2C%20and%20in%20silico%20analysis%20of%20wildtype%20protein%20and%20mutants%20in%20the%20presence%20or%20absence%20of%20an%20active-site%20inhibitor%2C%20we%20experimentally%20validate%20the%20obtained%20architecture%20with%20respect%20to%20global%20protein%20motion%20and%20long-range%20allosteric%20modulation.%20Notably%2C%20the%20identified%20network%20highlights%20communication%20pathways%20across%20several%20functional%20sites%2C%20prominently%20involving%20the%20allosteric%20site%2C%20the%20activation%20loop%2C%20and%20the%20lipid-binding%20domain%20with%20its%20embedded%20cryptic%20pocket%20in%20the%20C-lobe.%20These%20findings%20provide%20mechanistic%20insight%20into%20p38%5Cu03b1%20allostery%20and%20suggest%20new%20opportunities%20for%20the%20rational%20design%20of%20allosteric%20modulators%20of%20MAP%20kinases.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026-03-09%22%2C%22DOI%22%3A%2210.21203%5C%2Frs.3.rs-8958159%5C%2Fv1%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fwww.researchsquare.com%5C%2Farticle%5C%2Frs-8958159%5C%2Fv1%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T22%3A01%3A25Z%22%7D%7D%2C%7B%22key%22%3A%22XNK7CNRV%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Lewis%20and%20Constante-Amores%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BLewis%2C%20I.%20J.%20G.%20%26amp%3B%20Constante-Amores%2C%20C.%20R.%20Exact%20coherent%20states%20underlying%20chaotic%20falling-film%20dynamics.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.07297%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.07297%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Exact%20coherent%20states%20underlying%20chaotic%20falling-film%20dynamics%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Isaac%20J.%20G.%22%2C%22lastName%22%3A%22Lewis%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22C.%20Ricardo%22%2C%22lastName%22%3A%22Constante-Amores%22%7D%5D%2C%22abstractNote%22%3A%22Dynamical-systems%20approaches%20to%20spatiotemporal%20chaos%20have%20been%20developed%20primarily%20for%20single-phase%20flows%2C%20where%20the%20system%20state%20is%20defined%20by%20bulk%20velocity%20fields.%20Extending%20these%20ideas%20to%20two-phase%20flows%20remains%20challenging%20because%20the%20dynamics%20are%20intrinsically%20coupled%20to%20the%20evolution%20of%20a%20deforming%20interface.%20Here%2C%20we%20address%20this%20challenge%20for%20a%20two-dimensional%20vertical%20falling%20film%20by%20formulating%20the%20dynamics%20in%20terms%20of%20the%20interface%20evolution.%20Starting%20from%20the%20Navier--Stokes%20equations%2C%20we%20recover%20a%20classical%20long-wave%20interface%20evolution%20equation%2C%20originally%20derived%20by%20Topper%20%26amp%3B%20Kawahara%20%281978%29.%20Using%20this%20formulation%2C%20we%20perform%20an%20extensive%20parametric%20study%20to%20construct%20a%20regime%20map%20in%20the%20space%20of%20domain%20size%20and%20dispersion%20parameter.%20The%20resulting%20map%20reveals%20a%20rich%20range%20of%20interfacial%20behaviors%2C%20including%20travelling%20waves%2C%20bursting%20travelling%20waves%2C%20and%20fully%20chaotic%20regimes.%20In%20the%20chaotic%20falling%20film%20regime%2C%20we%20exploit%20the%20dissipative%20nature%20of%20the%20governing%20equation%2C%20which%20suggests%20that%20the%20long-time%20dynamics%20evolve%20onto%20an%20inertial%20manifold.%20Using%20a%20data-driven%20approach%2C%20we%20parameterize%20this%20inertial%20manifold%20and%20estimate%20its%20intrinsic%20dimension%2C%20suggesting%20approximately%20linear%20growth%20with%20domain%20size.%20We%20then%20construct%20low-dimensional%20models%20in%20manifold%20coordinates%20to%20facilitate%20the%20search%20for%20exact%20coherent%20states%20of%20the%20full%20system.%20Using%20this%20approach%2C%20we%20identify%20travelling%20waves%2C%20relative%20periodic%20orbits%20and%20equilibria%20embedded%20within%20the%20chaotic%20attractor.%20Chaotic%20trajectories%20repeatedly%20approach%20the%20neighbourhoods%20of%20these%20invariant%20solutions%2C%20indicating%20that%20the%20recurrent%20interfacial%20patterns%20observed%20in%20the%20dynamics%20correspond%20to%20visits%20to%20these%20coherent%20states.%20To%20the%20best%20of%20our%20knowledge%2C%20this%20constitutes%20the%20first%20identification%20of%20exact%20coherent%20structures%20embedded%20in%20chaotic%20falling-film%20dynamics.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.07297%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.07297%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A58%3A46Z%22%7D%7D%2C%7B%22key%22%3A%22FAA2SIDV%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Prasad%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BPrasad%2C%20V.%20Accelerated%20Time-domain%20Analysis%20for%20Gravitational%20Wave%20Astronomy.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.06010%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.06010%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Accelerated%20Time-domain%20Analysis%20for%20Gravitational%20Wave%20Astronomy%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Vaishak%22%2C%22lastName%22%3A%22Prasad%22%7D%5D%2C%22abstractNote%22%3A%22Most%20current%20compact-binary%20searches%20and%20parameter-estimation%20pipelines%20evaluate%20the%20Gaussian-noise%20likelihood%20approximately%20using%20frequency-domain%20inner%20products%20with%20great%20success%20in%20analyzing%20gravitational-wave%20signals.%20This%20is%20historically%20motivated%20by%20%28i%29%20the%20approximate%20stationarity%20of%20detector%20noise%20on%20sufficiently%20long%20timescales%2C%20allowing%20a%20circulant%20approximation%20in%20the%20domain%20that%20diagonalizes%20the%20noise%20covariance%20in%20the%20Fourier%20basis%2C%20and%20%28ii%29%20the%20efficiency%20of%20matched%20filtering%20via%20fast%20Fourier%20transforms.%20However%2C%20the%20advantage%20of%20frequency-domain%20analysis%20comes%20with%20its%20own%20limitations.%20In%20this%20article%2C%20we%20develop%20a%20self-contained%2C%20end-to-end%2C%20%5C%5Cemph%7Bfully%20time-domain%7D%20formulation%20of%20gravitational-wave%20inference%20and%20present%20an%20implementation%20that%20makes%20the%20likelihood%20evaluation%20practical%20at%20scale%20by%20exploiting%20structured%20linear%20algebra%2C%20software%2C%20and%20hardware%20acceleration.%20We%20validate%20the%20method%20using%20injections%20and%20demonstrate%20speedups%20for%20likelihood%20evaluation%20and%20on%20modern%20GPUs.%20We%20present%20%5C%5Cemph%7Btdanalysis%7D%2C%20an%20accelerated%20implementation%20that%20handles%20gaps%2C%20sharp%20boundaries%2C%20and%20multiple%20disjoint%20segments%2C%20and%20supports%20GPUs.%20We%20demonstrate%20some%20of%20its%20applications%20in%20gravitational%20wave%20astronomy.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.06010%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.06010%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A56%3A04Z%22%7D%7D%2C%7B%22key%22%3A%22FLJ7Q9UV%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Prasad%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BPrasad%2C%20V.%20Multi-Segment%20Consistency%20Tests%20of%20General%20Relativity.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.05835%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.05835%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Multi-Segment%20Consistency%20Tests%20of%20General%20Relativity%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Vaishak%22%2C%22lastName%22%3A%22Prasad%22%7D%5D%2C%22abstractNote%22%3A%22As%20the%20LIGO-VIRGO-KAGRA%20Network%20of%20gravitational-wave%20detectors%20improves%20in%20sensitivity%2C%20accumulating%20hundreds%20of%20gravitational-wave%20detections%20per%20year%2C%20it%20becomes%20imperative%20to%20improve%20tests%20of%20general%20relativity%20in%20concert.%20The%20test%20of%20Hawking%26%23039%3Bs%20law%20of%20area%20increase%20has%20gained%20prominence%20since%20GW250114%2C%20where%20black%20holes%20in%20General%20Relativity%20were%20tested%20with%20unprecedented%20precision%2C%20using%20the%20linear%20ringdown%20and%20pre-merger%20portions%20of%20the%20signal.%20A%20closely%20related%20test%20is%20the%20Inspiral-Merger-Ringdown%20Consistency%20Test%2C%20which%20assesses%20the%20consistency%20of%20the%20high-%20and%20low-frequency%20parts%20of%20the%20signals.%20In%20this%20letter%2C%20I%20present%20a%20multi-parameter%20Multi-Segment%20Consistency%20Test%20%28MSCT%29%20that%20generalizes%20and%20improves%20upon%20existing%20tests%20by%20ensuring%20that%20the%20extrinsic%20properties%20of%20the%20signal%20are%20consistent%20across%20its%20independent%20segments%20and%20by%20adopting%20an%20accelerated%20time-domain%20approach.%20The%20improved%20area%20law%20test%20is%20then%20presented%20as%20a%20projection%20of%20this%20MSCT%20test.%20These%20crucial%20improvements%2C%20which%20bring%20physical%20consistency%20to%20the%20area%20law%20test%2C%20lead%20to%20more%20stringent%20constraints%20on%20the%20increase%20in%20estimated%20area%20from%20observed%20binary%20black%20hole%20mergers%2C%20while%20also%20capturing%20covariances%20among%20the%20parameters.%5Cn%20Applying%20the%20two-segment%20version%20of%20this%20test%20to%20the%20inspiral%20and%20ringdown%20parts%20of%20GW250114%2C%20and%20keeping%20some%20of%20the%20extrinsic%20parameters%20common%20between%20the%20segments%2C%20I%20test%20the%20signal%20to%20unprecedented%20accuracy%2C%20obtaining%20%244.61%20%5E%7B%2B0.24%7D%20_%7B-0.11%7D%5Cu03c3%24%20significant%20result%20for%20the%20area%20increase%2C%20even%20as%20more%20than%204%20pre-merger%20cycles%20of%20the%20signal%20are%20excluded%20from%20the%20analysis.%20Also%2C%20I%20infer%20that%20the%20final%20state%20lies%20within%20the%2015%5C%5C%25%20highest%20posterior%20density%20confidence%20interval.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.05835%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.05835%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A47%3A08Z%22%7D%7D%2C%7B%22key%22%3A%22QIMT6U98%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Chen%20et%20al.%22%2C%22parsedDate%22%3A%222026%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BChen%2C%20J.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20Equilibrium%20Thermochemistry%20and%20Crystallographic%20Morphology%20of%20Manganese%20Sulfide%20Nanocrystals.%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.05420%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FARXIV.2603.05420%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22Equilibrium%20Thermochemistry%20and%20Crystallographic%20Morphology%20of%20Manganese%20Sulfide%20Nanocrystals%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Junchi%22%2C%22lastName%22%3A%22Chen%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Tamilarasan%22%2C%22lastName%22%3A%22Subramani%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Deep%22%2C%22lastName%22%3A%22Mekan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Danielle%22%2C%22lastName%22%3A%22Gendler%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Ray%22%2C%22lastName%22%3A%22Yang%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Manish%22%2C%22lastName%22%3A%22Kumar%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Megan%22%2C%22lastName%22%3A%22Householder%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Alexis%20Rosado%22%2C%22lastName%22%3A%22Ortiz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Emil%20A.%22%2C%22lastName%22%3A%22Hernandez-Pagan%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Kristina%22%2C%22lastName%22%3A%22Lilova%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Robert%20B.%22%2C%22lastName%22%3A%22Wexler%22%7D%5D%2C%22abstractNote%22%3A%22Manganese%20sulfide%20%28MnS%29%20is%20a%20p-type%20magnetic%20semiconductor%20whose%20physicochemical%20properties%20are%20sensitive%20to%20nanocrystal%20%28NC%29%20morphology%2C%20yet%20the%20thermodynamic%20driving%20forces%20governing%20morphology%20across%20MnS%20polymorphs%20remain%20poorly%20understood.%20Here%2C%20we%20use%20density%20functional%20theory%20%28DFT%29%20to%20predict%20the%20equilibrium%20morphologies%20of%20rock%20salt%20%28RS%29%2C%20zinc%20blende%20%28ZB%29%2C%20and%20wurtzite%20%28WZ%29%20MnS%20NCs%20as%20a%20function%20of%20the%20relative%20chemical%20potential%20of%20sulfur%2C%20%24%5Cu0394%5Cu03bc_%7BS%7D%24.%20Benchmarking%20against%20Heyd%24%5C%5Cunicode%7Bx2013%7D%24Scuseria%24%5C%5Cunicode%7Bx2013%7D%24Ernzerhof%20%28HSE06%29%20hybrid%20functional%20calculations%20reveals%20that%20the%20r%24%5E2%24SCAN%20meta-generalized%20gradient%20approximation%20reproduces%20experimental%20lattice%20constants%20and%20thermochemical%20reaction%20energies%20but%20underestimates%20S-terminated%20polar%20surface%20energies%20by%20up%20to%20a%20factor%20of%20five%3B%20applying%20a%20Hubbard%20%24U%24%20correction%20%28r%24%5E2%24SCAN%2B%24U%24%2C%20%24U%20%3D%202.7%24%20eV%29%20to%20the%20Mn%203d%20states%20brings%20the%20results%20into%20close%20agreement%20with%20HSE06.%20Using%20the%20validated%20r%24%5E2%24SCAN%2B%24U%24%20framework%20with%20the%20Gibbs%24%5C%5Cunicode%7Bx2013%7D%24Wulff%20theorem%2C%20we%20predict%20that%20RS-MnS%20NCs%20favor%20nanocubes%20across%20nearly%20the%20entire%20stability%20window%2C%20ZB-MnS%20NCs%20transform%20from%20rhombic%20dodecahedra%20%28Mn-rich%29%20to%20polyhedra%20with%2016%20triangular%20faces%20%28S-rich%29%2C%20and%20WZ-MnS%20NCs%20adopt%20rod-like%20morphologies%20with%20%24%5Cu0394%5Cu03bc_%7BS%7D%24-sensitive%20base%20truncation.%20Synthesized%20RS-MnS%20NCs%20confirm%20the%20predicted%20cubic%20morphology%2C%20and%20high-temperature%20oxidative%20solution%20calorimetry%20yields%20an%20apparent%20surface%20energy%20of%201.15%20%24%5C%5Cpm%24%200.38%20J%24%5C%5Ccdot%24m%24%5E%7B-2%7D%24%2C%20higher%20than%20the%20theoretical%20equilibrium%20value%20%280.42%24%5C%5Cunicode%7Bx2013%7D%240.43%20J%24%5C%5Ccdot%24m%24%5E%7B-2%7D%24%29%20due%20to%20high-index%20facet%20exposure%2C%20surface%20area%20uncertainty%2C%20and%20non-ideal%20surface%20configurations%20in%20real%20samples.%20This%20work%20establishes%20a%20framework%20for%20predicting%20the%20equilibrium%20morphologies%20of%20metal%20chalcogenide%20NCs.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22%22%2C%22date%22%3A%222026%22%2C%22DOI%22%3A%2210.48550%5C%2FARXIV.2603.05420%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.05420%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A42%3A04Z%22%7D%7D%2C%7B%22key%22%3A%226ZJ5RW5I%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Agrawalla%20et%20al.%22%2C%22parsedDate%22%3A%222026-03-04%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BAgrawalla%2C%20B.%2C%20Nauman%2C%20M.%20%26amp%3B%20Kumar%2C%20A.%20What%20Does%20Flow%20Matching%20Bring%20To%20TD%20Learning%3F%20Preprint%20at%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2603.04333%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2603.04333%26lt%3B%5C%2Fa%26gt%3B%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22preprint%22%2C%22title%22%3A%22What%20Does%20Flow%20Matching%20Bring%20To%20TD%20Learning%3F%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Bhavya%22%2C%22lastName%22%3A%22Agrawalla%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michal%22%2C%22lastName%22%3A%22Nauman%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aviral%22%2C%22lastName%22%3A%22Kumar%22%7D%5D%2C%22abstractNote%22%3A%22Recent%20work%20shows%20that%20flow%20matching%20can%20be%20effective%20for%20scalar%20Q-value%20function%20estimation%20in%20reinforcement%20learning%20%28RL%29%2C%20but%20it%20remains%20unclear%20why%20or%20how%20this%20approach%20differs%20from%20standard%20critics.%20Contrary%20to%20conventional%20belief%2C%20we%20show%20that%20their%20success%20is%20not%20explained%20by%20distributional%20RL%2C%20as%20explicitly%20modeling%20return%20distributions%20can%20reduce%20performance.%20Instead%2C%20we%20argue%20that%20the%20use%20of%20integration%20for%20reading%20out%20values%20and%20dense%20velocity%20supervision%20at%20each%20step%20of%20this%20integration%20process%20for%20training%20improves%20TD%20learning%20via%20two%20mechanisms.%20First%2C%20it%20enables%20robust%20value%20prediction%20through%20%5C%5Cemph%7Btest-time%20recovery%7D%2C%20whereby%20iterative%20computation%20through%20integration%20dampens%20errors%20in%20early%20value%20estimates%20as%20more%20integration%20steps%20are%20performed.%20This%20recovery%20mechanism%20is%20absent%20in%20monolithic%20critics.%20Second%2C%20supervising%20the%20velocity%20field%20at%20multiple%20interpolant%20values%20induces%20more%20%5C%5Cemph%7Bplastic%7D%20feature%20learning%20within%20the%20network%2C%20allowing%20critics%20to%20represent%20non-stationary%20TD%20targets%20without%20discarding%20previously%20learned%20features%20or%20overfitting%20to%20individual%20TD%20targets%20encountered%20during%20training.%20We%20formalize%20these%20effects%20and%20validate%20them%20empirically%2C%20showing%20that%20flow-matching%20critics%20substantially%20outperform%20monolithic%20critics%20%282%24%5C%5Ctimes%24%20in%20final%20performance%20and%20around%205%24%5C%5Ctimes%24%20in%20sample%20efficiency%29%20in%20settings%20where%20loss%20of%20plasticity%20poses%20a%20challenge%20e.g.%2C%20in%20high-UTD%20online%20RL%20problems%2C%20while%20remaining%20stable%20during%20learning.%22%2C%22genre%22%3A%22%22%2C%22repository%22%3A%22arXiv%22%2C%22archiveID%22%3A%22arXiv%3A2603.04333%22%2C%22date%22%3A%222026-03-04%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2603.04333%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22http%3A%5C%2F%5C%2Farxiv.org%5C%2Fabs%5C%2F2603.04333%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A29%3A30Z%22%7D%7D%2C%7B%22key%22%3A%225S3C9DZT%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Hill%20et%20al.%22%2C%22parsedDate%22%3A%222026-03-20%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BHill%2C%20A.%20F.%2C%20Ruiz-Escudero%2C%20A.%20%26amp%3B%20Montemore%2C%20M.%20M.%20Few-Shot%20Ensemble%20Learning%20for%20Catalysis%20and%20Application%20to%20Trimetallics%20for%20Oxygen%20Reduction.%20%26lt%3Bi%26gt%3BACS%20Catal.%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Bb%26gt%3B16%26lt%3B%5C%2Fb%26gt%3B%2C%205592%26%23x2013%3B5602%20%282026%29.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Few-Shot%20Ensemble%20Learning%20for%20Catalysis%20and%20Application%20to%20Trimetallics%20for%20Oxygen%20Reduction%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Avery%20F.%22%2C%22lastName%22%3A%22Hill%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Andrea%22%2C%22lastName%22%3A%22Ruiz-Escudero%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Matthew%20M.%22%2C%22lastName%22%3A%22Montemore%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22date%22%3A%222026-03-20%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1021%5C%2Facscatal.5c08168%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fpubs.acs.org%5C%2Fdoi%5C%2F10.1021%5C%2Facscatal.5c08168%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222155-5435%2C%202155-5435%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A26%3A41Z%22%7D%7D%2C%7B%22key%22%3A%22UG8QHR6H%22%2C%22library%22%3A%7B%22id%22%3A5005740%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Yoo%20et%20al.%22%2C%22parsedDate%22%3A%222026-02-24%22%2C%22numChildren%22%3A0%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%20style%3D%26quot%3Bclear%3A%20left%3B%20%26quot%3B%26gt%3B%5Cn%20%20%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-left-margin%26quot%3B%20style%3D%26quot%3Bfloat%3A%20left%3B%20padding-right%3A%200.5em%3B%20text-align%3A%20right%3B%20width%3A%201em%3B%26quot%3B%26gt%3B1.%26lt%3B%5C%2Fdiv%26gt%3B%26lt%3Bdiv%20class%3D%26quot%3Bcsl-right-inline%26quot%3B%20style%3D%26quot%3Bmargin%3A%200%20.4em%200%201.5em%3B%26quot%3B%26gt%3BYoo%2C%20J.%20Y.%20%26lt%3Bi%26gt%3Bet%20al.%26lt%3B%5C%2Fi%26gt%3B%20In%20Silico%20Reconstruction%20of%20Primary%20and%20Metastatic%20Tumor%20Architecture%20using%20Geographic%20Information%20System-Augmented%20Spatial%20Transcriptomics.%20%26lt%3Bi%26gt%3BCancer%20Research%26lt%3B%5C%2Fi%26gt%3B%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1158%5C%2F0008-5472.CAN-25-3161%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1158%5C%2F0008-5472.CAN-25-3161%26lt%3B%5C%2Fa%26gt%3B%20%282026%29%20doi%3A10.1158%5C%2F0008-5472.CAN-25-3161.%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%20%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22In%20Silico%20Reconstruction%20of%20Primary%20and%20Metastatic%20Tumor%20Architecture%20using%20Geographic%20Information%20System-Augmented%20Spatial%20Transcriptomics%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Jin%20Young%22%2C%22lastName%22%3A%22Yoo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Sabrina%22%2C%22lastName%22%3A%22Akter%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Qianying%22%2C%22lastName%22%3A%22Zuo%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Matthew%22%2C%22lastName%22%3A%22Rilloraza%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Amir%22%2C%22lastName%22%3A%22Kazemi%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Wenjun%22%2C%22lastName%22%3A%22Wu%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Mahima%22%2C%22lastName%22%3A%22Goel%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Audrey%22%2C%22lastName%22%3A%22Lam%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Debapriya%22%2C%22lastName%22%3A%22Dutta%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Betsy%22%2C%22lastName%22%3A%22Barnick%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Melike%22%2C%22lastName%22%3A%22Pekmezci%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Maria%22%2C%22lastName%22%3A%22Grosse%20Perdekamp%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Aiman%22%2C%22lastName%22%3A%22Soliman%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Zeynep%22%2C%22lastName%22%3A%22Madak-Erdogan%22%7D%5D%2C%22abstractNote%22%3A%22Abstract%20%5Cn%20%20%20%20%20%20%20%20%20%20%20%20The%20tumor%20microenvironment%20comprises%20different%20cell%20populations%20that%20interact%2C%20contributing%20to%20tumor%20heterogeneity%20and%20therapy%20response.%20Spatial%20transcriptomics%20offers%20valuable%20insights%20into%20transcriptional%20complexity%20and%20heterogeneity%20of%20the%20tumor%20microenvironment.%20Here%2C%20we%20established%20geographic%20information%20system%20%28GIS%29-augmented%20in%20silico%20reconstruction%20of%20tumor%20architecture%20%28GIS-ROTA%29%2C%20a%20biologically%20informed%20analytical%20framework%20that%20integrated%20pathway%20or%20cell%20type-based%20enrichment%20analysis%20with%20spatial%20autocorrelation%20measurement%20to%20uncover%20functional%20spatial%20domains.%20The%20approach%20considered%20biological%20functions%20prior%20to%20identifying%20any%20spatial%20domains%2C%20providing%20direct%20interpretability%20and%20minimizing%20the%20subjectivity%20of%20interpreting%20clusters%20observed%20from%20conventional%20analytical%20methods.%20Application%20of%20GIS-ROTA%20to%20a%20Visium%20spatial%20transcriptomics%20dataset%20of%20primary%20and%20metastatic%20estrogen%20receptor-positive%20breast%20tumor%20samples%20revealed%20extensive%20co-localization%20of%20estrogen%20response%20with%20metabolic%20pathway%20gene%20sets%20and%20mutual%20exclusivity%20with%20metastasis-related%20and%20specific%20immune-related%20pathway%20gene%20sets.%20Overall%2C%20the%20GIS-ROTA%20framework%20integrates%20biological%20knowledge%20first%2C%20yielding%20spatial%20patterns%20with%20functional%20relevance%20and%20enabling%20identification%20of%20novel%20targets%20for%20development%20of%20therapeutic%20strategies.%22%2C%22date%22%3A%222026-02-24%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.1158%5C%2F0008-5472.CAN-25-3161%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Faacrjournals.org%5C%2Fcancerres%5C%2Farticle%5C%2Fdoi%5C%2F10.1158%5C%2F0008-5472.CAN-25-3161%5C%2F774840%5C%2FIn-Silico-Reconstruction-of-Primary-and-Metastatic%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%220008-5472%2C%201538-7445%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%22XHSH9DGT%22%5D%2C%22dateModified%22%3A%222026-04-20T21%3A20%3A03Z%22%7D%7D%5D%7D
1.
Hsu, H.-Y., Cheng, T., Wen, J., Schwing, A. G. & Wang, S. Seeing Without Eyes: 4D Human-Scene Understanding from Wearable IMUs. Preprint at https://doi.org/10.48550/ARXIV.2604.21926 (2026).
1.
Wang, D., Zhang, Y., Qin, Y., Shao, Q. & Xu, D. Efficient Adaptation of Structure‐Aware Protein Language Models for Diverse Protein Applications. Current Protocols 6, e70377 (2026).
1.
Lee, W. et al. Core/Shell HgCdTe/HgCdSe Quantum Dots for Wavefunction Engineering with Infrared Band Gaps. Preprint at https://doi.org/10.26434/chemrxiv.15002806/v1 (2026).
1.
Mehdi, S., Cho, I. & Isayev, O. Knowing when to trust machine-learned interatomic potentials. Preprint at https://doi.org/10.48550/ARXIV.2605.00640 (2026).
1.
Chan, J. et al. ReLay: Personalized LLM-Generated Plain-Language Summaries for Better Understanding, but at What Cost? arXiv.org https://arxiv.org/abs/2605.00468v1 (2026).
1.
Yarici, Y. & AlRegib, G. MER-DG: Modality-Entropy Regularization for Multimodal Domain Generalization. arXiv.org https://arxiv.org/abs/2605.01967v1 (2026).
1.
Han, D., Shepherd, C., Benton, M. L. & Nordman, J. T. Nanopore-based sequencing of active DNA replication reveals key principles of metazoan replication dynamics. Sci. Adv. 12, eaed2806 (2026).
1.
Ma, Y. & Tuninetti, D. On Coded Caching Systems with Decentralized Linear Coding Placement. arXiv.org https://arxiv.org/abs/2604.27073v1 (2026).
1.
Wu, J. et al. Eccentricity as a signature of hierarchical subsolar-mass mergers in collapsar disks. arXiv.org https://arxiv.org/abs/2604.26912v1 (2026).
1.
Wang, B., Xu, G., Tang, Z., Zhang, J. & Cheng, Z. OmniShotCut: Holistic Relational Shot Boundary Detection with Shot-Query Transformer. arXiv.org https://arxiv.org/abs/2604.24762v1 (2026).
1.
Gao, R., Jeon, Y., Roy, S., Ziyadi, M. & Chen, X. ‘Anthony’. C-MORAL: Controllable Multi-Objective Molecular Optimization with Reinforcement Alignment for LLMs. arXiv.org https://arxiv.org/abs/2604.23061v1 (2026).
1.
Najafi, A. R. et al. Characterization and Computational Engineering of Structural Elements Controlling Gas Permeability in PIP2 ;1 Aquaporins. J Comput Chem 47, e70377 (2026).
1.
Kalyanasundaram, S. & Chaspari, T. Conditional Diffusion Models for Mental Health-Preserving Voice Conversion. in ICASSP 2026 - 2026 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) 17982–17986 (IEEE, Barcelona, Spain, 2026). http://doi.org/10.1109/ICASSP55912.2026.11464847.
1.
Howes, W., Ahmed, F., Kobayashi, K., Chakraborty, S. & Alam, S. B. Neuroscience Inspired Graph Operators Towards Edge-Deployable Virtual Sensing for Irregular Geometries. Preprint at https://doi.org/10.48550/arXiv.2604.16722 (2026).
1.
Dong, P., Swerdlow, A., Sadigh, D. & Finn, C. FASTER: Value-Guided Sampling for Fast RL. Preprint at https://doi.org/10.48550/ARXIV.2604.19730 (2026).
1.
Kaushal, N. & Penmatsa, A. M. FragBERTa: Exploring Fragment-based Molecular Representation Learning with SAFE. Preprint at https://doi.org/10.26434/chemrxiv.15000476/v2 (2026).
1.
Bouckley, E. R., Lewin, S. F. & Lefauve, A. Early onset of secondary shear instability in Kelvin-Helmholtz braids at high Reynolds number. Preprint at https://doi.org/10.48550/ARXIV.2604.16173 (2026).
1.
Bush, M., Palmer, J. C. & Howard, M. P. Simulating hydrodynamic interactions in colloidal suspensions using multiparticle collision dynamics with rigid-body constraints. Preprint at https://doi.org/10.48550/ARXIV.2604.14307 (2026).
1.
Roy, S., Chakraborty, S. & Alam, S. B. Beyond Uniform Sampling: Synergistic Active Learning and Input Denoising for Robust Neural Operators. Preprint at https://doi.org/10.48550/ARXIV.2604.13316 (2026).
1.
Motta, P. N., Neto, M. R., Prather, C. & Cárdenas-Avendaño, A. Sensitivities of Black Hole Images from GRMHD Simulations. Preprint at https://doi.org/10.48550/ARXIV.2604.11869 (2026).
1.
Shivaiah, K.-K. et al. A Conserved Sequence Insert of Fibrillins is Involved in Plastoglobule Association and Lipid Binding. Plant and Cell Physiology pcag047 (2026) http://doi.org/10.1093/pcp/pcag047.
1.
Naethe Motta, P., Nemmen, R. & Joshi, A. V. GPUmonty : A GPU-accelerated Relativistic Monte Carlo Radiative Transfer Code. ApJ 1001, 148 (2026).
1.
Gu, L., Huang, H.-J., Qadri, M., Kaess, M. & Yuan, W. TouchAnything: Diffusion-Guided 3D Reconstruction from Sparse Robot Touches. Preprint at https://doi.org/10.48550/ARXIV.2604.08945 (2026).
1.
Liu, E. et al. What do Language Models Learn and When? The Implicit Curriculum Hypothesis. Preprint at https://doi.org/10.48550/ARXIV.2604.08510 (2026).
1.
Jayakody, S., Zhao, Y., Nehate, C. D. & Wang, J. GhostServe: A Lightweight Checkpointing System in the Shadow for Fault-Tolerant LLM Serving. Preprint at https://doi.org/10.48550/ARXIV.2605.00831 (2026).
1.
Mo, C., Karsin, B., Adams, P. & Zhang, M. VecFlow-Chamfer: A GPU-based Data Management System for High-Performance Multi-Vector Search on Superchips. Proc. ACM Manag. Data 4, 1–26 (2026).
1.
Jiang, Y. et al. DeepDISC-Euclid: Source Classification and Photometric Redshifts in Euclid Deep Field North With a Pixel-Level Deep Learning Approach. Preprint at https://arxiv.org/abs/2604.03182v2 (2026).
1.
Howes, W. et al. Graph Neural Operator Towards Edge Deployability and Portability for Sparse-to-Dense, Real-Time Virtual Sensing on Irregular Grids. Preprint at https://doi.org/10.48550/ARXIV.2604.01802 (2026).
1.
Long, D., Zhe, S., Williams, S., Oliker, L. & Bai, Z. StFT: Spatio-temporal Fourier Transformer for Long-term Dynamics Prediction. Preprint at https://doi.org/10.48550/arXiv.2503.11899 (2025).
1.
Zhang, Z. et al. HandX: Scaling Bimanual Motion and Interaction Generation. Preprint at https://doi.org/10.48550/arXiv.2603.28766 (2026).
1.
Guo, J. & Schwalbe-Koda, D. Generative Inversion of Spectroscopic Data for Amorphous Structure Elucidation. Preprint at https://doi.org/10.48550/ARXIV.2603.23210 (2026).
1.
Cruz-Camacho, N., Conde-Ocazionez, C., Dexheimer, V., Noronha-Hostler, J. & Yunes, N. Sensitivity of neutron star observables to microscopic nuclear parameters of realistic equations of state. Preprint at https://doi.org/10.48550/ARXIV.2603.16019 (2026).
1.
Lihan, M., Pant, S., Ojoawo, A. M., Rienstra, C. M. & Tajkhorshid, E. Probing Solution Dynamics of Tissue Factor Using Molecular Dynamics Simulations Guided by NMR Chemical Shifts. J. Phys. Chem. B 130, 4528–4537 (2026).
1.
Li, C., Merkel, M. & Sussman, D. M. Models of 3D confluent tissue as under-constrained glasses. Preprint at https://doi.org/10.48550/ARXIV.2603.24438 (2026).
1.
Cassar, N. et al. $B \to π$, $B_{(s)} \to D_{(s)}$ from 2+1+1 Flavor Lattice QCD. Preprint at https://doi.org/10.48550/ARXIV.2603.23702 (2026).
1.
Lin, S. et al. D$_4$CNN$\times$AnaCal: Physics-Informed Machine Learning for Accurate and Precise Weak Lensing Shear Estimation. Preprint at https://doi.org/10.48550/ARXIV.2603.19046 (2026).
1.
Lazarian, A., Pogosyan, D. & Hu, Y. Model of super-Alfvénic MHD turbulence and structure functions of polarization. Preprint at https://doi.org/10.48550/ARXIV.2511.08800 (2025).
1.
Li, X. et al. MCLR: Improving Conditional Modeling via Inter-Class Likelihood-Ratio Maximization and Unifying Classifier-Free Guidance with Alignment Objectives. Preprint at https://doi.org/10.48550/ARXIV.2603.22364 (2026).
1.
Wang, S., Kang, W. & Li, C. A Tug-of-War Between Baroclinic Eddies and Convection: Implications for Icy Moon Oceans. Preprint at https://doi.org/10.48550/ARXIV.2603.17185 (2026).
1.
Meher, S. Build, Borrow, or Just Fine-Tune? A Political Scientist’s Guide to Choosing NLP Models. Preprint at https://doi.org/10.48550/ARXIV.2603.09595 (2026).
1.
Ung, S. F., Mahajan, A. & Reichman, D. R. Study of the triangular-lattice Hubbard model with constrained-path quantum Monte Carlo. Preprint at https://doi.org/10.48550/ARXIV.2603.14808 (2026).
1.
Wong, N. & Yang, J. H. Bias in Universal Machine-Learned Interatomic Potentials and its Effects on Fine-Tuning. Preprint at https://doi.org/10.48550/ARXIV.2603.10159 (2026).
1.
Linser, R., Acharyya, S., Weisner, J. & Bernardi, R. Mapping of dynamic allostery within p38 alpha kinase via network analyses and NMR spectroscopy. Preprint at https://doi.org/10.21203/rs.3.rs-8958159/v1 (2026).
1.
Lewis, I. J. G. & Constante-Amores, C. R. Exact coherent states underlying chaotic falling-film dynamics. Preprint at https://doi.org/10.48550/ARXIV.2603.07297 (2026).
1.
Prasad, V. Accelerated Time-domain Analysis for Gravitational Wave Astronomy. Preprint at https://doi.org/10.48550/ARXIV.2603.06010 (2026).
1.
Prasad, V. Multi-Segment Consistency Tests of General Relativity. Preprint at https://doi.org/10.48550/ARXIV.2603.05835 (2026).
1.
Chen, J. et al. Equilibrium Thermochemistry and Crystallographic Morphology of Manganese Sulfide Nanocrystals. Preprint at https://doi.org/10.48550/ARXIV.2603.05420 (2026).
1.
Agrawalla, B., Nauman, M. & Kumar, A. What Does Flow Matching Bring To TD Learning? Preprint at https://doi.org/10.48550/arXiv.2603.04333 (2026).
1.
Hill, A. F., Ruiz-Escudero, A. & Montemore, M. M. Few-Shot Ensemble Learning for Catalysis and Application to Trimetallics for Oxygen Reduction. ACS Catal. 16, 5592–5602 (2026).
1.
Yoo, J. Y. et al. In Silico Reconstruction of Primary and Metastatic Tumor Architecture using Geographic Information System-Augmented Spatial Transcriptomics. Cancer Research https://doi.org/10.1158/0008-5472.CAN-25-3161 (2026) doi:10.1158/0008-5472.CAN-25-3161.